Abstract
The normalization constraint on probability density poses a significant challenge for solving the Fokker-Planck equation. Normalizing Flow, an invertible generative model leverages the change of variables formula to ensure probability density conservation and enable the learning of complex data distributions. In this paper, we introduce Physics-Informed Normalizing Flows (PINF), a novel extension of continuous normalizing flows, incorporating diffusion through the method of characteristics. Our method, which is mesh-free and causality-free, can efficiently solve high dimensional time-dependent and steady-state Fokker-Planck equations.
A Review on AI Algorithms for Energy Management in E-Mobility Services
Authors: Sen Yan, Maqsood Hussain Shah, Ji Li, Noel O'Connor, Mingming Liu
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Systems and Control (eess.SY)
Abstract
E-mobility, or electric mobility, has emerged as a pivotal solution to address pressing environmental and sustainability concerns in the transportation sector. The depletion of fossil fuels, escalating greenhouse gas emissions, and the imperative to combat climate change underscore the significance of transitioning to electric vehicles (EVs). This paper seeks to explore the potential of artificial intelligence (AI) in addressing various challenges related to effective energy management in e-mobility systems (EMS). These challenges encompass critical factors such as range anxiety, charge rate optimization, and the longevity of energy storage in EVs. By analyzing existing literature, we delve into the role that AI can play in tackling these challenges and enabling efficient energy management in EMS. Our objectives are twofold: to provide an overview of the current state-of-the-art in this research domain and propose effective avenues for future investigations. Through this analysis, we aim to contribute to the advancement of sustainable and efficient e-mobility solutions, shaping a greener and more sustainable future for transportation.
3D Reconstruction with Generalizable Neural Fields using Scene Priors
Authors: Yang Fu, Shalini De Mello, Xueting Li, Amey Kulkarni, Jan Kautz, Xiaolong Wang, Sifei Liu
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Abstract
High-fidelity 3D scene reconstruction has been substantially advanced by recent progress in neural fields. However, most existing methods train a separate network from scratch for each individual scene. This is not scalable, inefficient, and unable to yield good results given limited views. While learning-based multi-view stereo methods alleviate this issue to some extent, their multi-view setting makes it less flexible to scale up and to broad applications. Instead, we introduce training generalizable Neural Fields incorporating scene Priors (NFPs). The NFP network maps any single-view RGB-D image into signed distance and radiance values. A complete scene can be reconstructed by merging individual frames in the volumetric space WITHOUT a fusion module, which provides better flexibility. The scene priors can be trained on large-scale datasets, allowing for fast adaptation to the reconstruction of a new scene with fewer views. NFP not only demonstrates SOTA scene reconstruction performance and efficiency, but it also supports single-image novel-view synthesis, which is underexplored in neural fields. More qualitative results are available at: https://oasisyang.github.io/neural-prior
Auto-grading C programming assignments with CodeBERT and Random Forest Regressor
Authors: Roshan Vasu Muddaluru, Sharvaani Ravikumar Thoguluva, Shruti Prabha, Dr. Peeta Basa Pati, Ms. Roshni M Balakrishnan
Abstract
Grading coding assignments manually is challenging due to complexity and subjectivity. However, auto-grading with deep learning simplifies the task. It objectively assesses code quality, detects errors, and assigns marks accurately, reducing the burden on instructors while ensuring efficient and fair assessment. This study provides an analysis of auto-grading of the C programming assignments using machine learning and deep learning approaches like regression, convolutional neural networks (CNN) and long short-term memory (LSTM). Using a code-based transformer word embedding model called CodeBERT, the textual code inputs were transformed into vectors, and the vectors were then fed into several models. The testing findings demonstrated the efficacy of the suggested strategy with a root mean squared error (RMSE) of 1.89. The contrast between statistical methods and deep learning techniques is discussed in the study.
VPA: Fully Test-Time Visual Prompt Adaptation
Authors: Jiachen Sun, Mark Ibrahim, Melissa Hall, Ivan Evtimov, Z. Morley Mao, Cristian Canton Ferrer, Caner Hazirbas
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Abstract
Textual prompt tuning has demonstrated significant performance improvements in adapting natural language processing models to a variety of downstream tasks by treating hand-engineered prompts as trainable parameters. Inspired by the success of textual prompting, several studies have investigated the efficacy of visual prompt tuning. In this work, we present Visual Prompt Adaptation (VPA), the first framework that generalizes visual prompting with test-time adaptation. VPA introduces a small number of learnable tokens, enabling fully test-time and storage-efficient adaptation without necessitating source-domain information. We examine our VPA design under diverse adaptation settings, encompassing single-image, batched-image, and pseudo-label adaptation. We evaluate VPA on multiple tasks, including out-of-distribution (OOD) generalization, corruption robustness, and domain adaptation. Experimental results reveal that VPA effectively enhances OOD generalization by 3.3% across various models, surpassing previous test-time approaches. Furthermore, we show that VPA improves corruption robustness by 6.5% compared to strong baselines. Finally, we demonstrate that VPA also boosts domain adaptation performance by relatively 5.2%. Our VPA also exhibits marked effectiveness in improving the robustness of zero-shot recognition for vision-language models.
Memory-Efficient Continual Learning Object Segmentation for Long Video
Authors: Amir Nazemi, Mohammad Javad Shafiee, Zahra Gharaee, Paul Fieguth
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Abstract
Recent state-of-the-art semi-supervised Video Object Segmentation (VOS) methods have shown significant improvements in target object segmentation accuracy when information from preceding frames is used in undertaking segmentation on the current frame. In particular, such memory-based approaches can help a model to more effectively handle appearance changes (representation drift) or occlusions. Ideally, for maximum performance, online VOS methods would need all or most of the preceding frames (or their extracted information) to be stored in memory and be used for online learning in consecutive frames. Such a solution is not feasible for long videos, as the required memory size would grow without bound. On the other hand, these methods can fail when memory is limited and a target object experiences repeated representation drifts throughout a video. We propose two novel techniques to reduce the memory requirement of online VOS methods while improving modeling accuracy and generalization on long videos. Motivated by the success of continual learning techniques in preserving previously-learned knowledge, here we propose Gated-Regularizer Continual Learning (GRCL), which improves the performance of any online VOS subject to limited memory, and a Reconstruction-based Memory Selection Continual Learning (RMSCL) which empowers online VOS methods to efficiently benefit from stored information in memory. Experimental results show that the proposed methods improve the performance of online VOS models up to 10 %, and boosts their robustness on long-video datasets while maintaining comparable performance on short-video datasets DAVIS16 and DAVIS17.
Efficient Low-rank Backpropagation for Vision Transformer Adaptation
Abstract
The increasing scale of vision transformers (ViT) has made the efficient fine-tuning of these large models for specific needs a significant challenge in various applications. This issue originates from the computationally demanding matrix multiplications required during the backpropagation process through linear layers in ViT. In this paper, we tackle this problem by proposing a new Low-rank BackPropagation via Walsh-Hadamard Transformation (LBP-WHT) method. Intuitively, LBP-WHT projects the gradient into a low-rank space and carries out backpropagation. This approach substantially reduces the computation needed for adapting ViT, as matrix multiplication in the low-rank space is far less resource-intensive. We conduct extensive experiments with different models (ViT, hybrid convolution-ViT model) on multiple datasets to demonstrate the effectiveness of our method. For instance, when adapting an EfficientFormer-L1 model on CIFAR100, our LBP-WHT achieves 10.4% higher accuracy than the state-of-the-art baseline, while requiring 9 MFLOPs less computation. As the first work to accelerate ViT adaptation with low-rank backpropagation, our LBP-WHT method is complementary to many prior efforts and can be combined with them for better performance.
joint prediction and denoising for large-scale multilingual self-supervised learning
Authors: William Chen, Jiatong Shi, Brian Yan, Dan Berrebbi, Wangyou Zhang, Yifan Peng, Xuankai Chang, Soumi Maiti, Shinji Watanabe
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Sound (cs.SD); Audio and Speech Processing (eess.AS)
Abstract
Multilingual self-supervised learning (SSL) has often lagged behind state-of-the-art (SOTA) methods due to the expenses and complexity required to handle many languages. This further harms the reproducibility of SSL, which is already limited to few research groups due to its resource usage. We show that more powerful techniques can actually lead to more efficient pre-training, opening SSL to more research groups. We propose WavLabLM, which extends WavLM's joint prediction and denoising to 40k hours of data across 136 languages. To build WavLabLM, we devise a novel multi-stage pre-training method, designed to address the language imbalance of multilingual data. WavLabLM achieves comparable performance to XLS-R on ML-SUPERB with less than 10% of the training data, making SSL realizable with academic compute. We show that further efficiency can be achieved with a vanilla HuBERT Base model, which can maintain 94% of XLS-R's performance with only 3% of the data, 4 GPUs, and limited trials. We open-source all code and models in ESPnet.
Beyond the Chat: Executable and Verifiable Text-Editing with LLMs
Authors: Philippe Laban, Jesse Vig, Marti A. Hearst, Caiming Xiong, Chien-Sheng Wu
Subjects: Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
Abstract
Conversational interfaces powered by Large Language Models (LLMs) have recently become a popular way to obtain feedback during document editing. However, standard chat-based conversational interfaces do not support transparency and verifiability of the editing changes that they suggest. To give the author more agency when editing with an LLM, we present InkSync, an editing interface that suggests executable edits directly within the document being edited. Because LLMs are known to introduce factual errors, Inksync also supports a 3-stage approach to mitigate this risk: Warn authors when a suggested edit introduces new information, help authors Verify the new information's accuracy through external search, and allow an auditor to perform an a-posteriori verification by Auditing the document via a trace of all auto-generated content. Two usability studies confirm the effectiveness of InkSync's components when compared to standard LLM-based chat interfaces, leading to more accurate, more efficient editing, and improved user experience.
Multi-dimensional Data Quick Query for Blockchain-based Federated Learning
Abstract
Due to the drawbacks of Federated Learning (FL) such as vulnerability of a single central server, centralized federated learning is shifting to decentralized federated learning, a paradigm which takes the advantages of blockchain. A key enabler for adoption of blockchain-based federated learning is how to select suitable participants to train models collaboratively. Selecting participants by storing and querying the metadata of data owners on blockchain could ensure the reliability of selected data owners, which is helpful to obtain high-quality models in FL. However, querying multi-dimensional metadata on blockchain needs to traverse every transaction in each block, making the query time-consuming. An efficient query method for multi-dimensional metadata in the blockchain for selecting participants in FL is absent and challenging. In this paper, we propose a novel data structure to improve the query efficiency within each block named MerkleRB-Tree. In detail, we leverage Minimal Bounding Rectangle(MBR) and bloom-filters for the query process of multi-dimensional continuous-valued attributes and discrete-valued attributes respectively. Furthermore, we migrate the idea of the skip list along with an MBR and a bloom filter at the head of each block to enhance the query efficiency for inter-blocks. The performance analysis and extensive evaluation results on the benchmark dataset demonstrate the superiority of our method in blockchain-based FL.
Human-robot Matching and Routing for Multi-robot Tour Guiding under Time Uncertainty
Authors: Bo Fu, Tribhi Kathuria, Denise Rizzo, Matthew Castanier, X. Jessie Yang, Maani Ghaffari, Kira Barton
Subjects: Robotics (cs.RO); Multiagent Systems (cs.MA)
Abstract
This work presents a framework for multi-robot tour guidance in a partially known environment with uncertainty, such as a museum. A simultaneous matching and routing problem (SMRP) is formulated to match the humans with robot guides according to their requested places of interest (POIs) and generate the routes for the robots according to uncertain time estimation. A large neighborhood search algorithm is developed to efficiently find sub-optimal low-cost solutions for the SMRP. The scalability and optimality of the multi-robot planner are evaluated computationally. The largest case tested involves 50 robots, 250 humans, and 50 POIs. A photo-realistic multi-robot simulation was developed to verify the tour guiding performance in an uncertain indoor environment.
PPG to ECG Signal Translation for Continuous Atrial Fibrillation Detection via Attention-based Deep State-Space Modeling
Abstract
An electrocardiogram (ECG or EKG) is a medical test that measures the heart's electrical activity. ECGs are often used to diagnose and monitor a wide range of heart conditions, including arrhythmias, heart attacks, and heart failure. On the one hand, the conventional ECG requires clinical measurement, which restricts its deployment to medical facilities. On the other hand, single-lead ECG has become popular on wearable devices using administered procedures. An alternative to ECG is Photoplethysmography (PPG), which uses non-invasive, low-cost optical methods to measure cardiac physiology, making it a suitable option for capturing vital heart signs in daily life. As a result, it has become increasingly popular in health monitoring and is used in various clinical and commercial wearable devices. While ECG and PPG correlate strongly, the latter does not offer significant clinical diagnostic value. Here, we propose a subject-independent attention-based deep state-space model to translate PPG signals to corresponding ECG waveforms. The model is highly data-efficient by incorporating prior knowledge in terms of probabilistic graphical models. Notably, the model enables the detection of atrial fibrillation (AFib), the most common heart rhythm disorder in adults, by complementing ECG's accuracy with continuous PPG monitoring. We evaluated the model on 55 subjects from the MIMIC III database. Quantitative and qualitative experimental results demonstrate the effectiveness and efficiency of our approach.
Adversarial Object Rearrangement in Constrained Environments with Heterogeneous Graph Neural Networks
Authors: Xibai Lou, Houjian Yu, Ross Worobel, Yang Yang, Changhyun Choi
Abstract
Adversarial object rearrangement in the real world (e.g., previously unseen or oversized items in kitchens and stores) could benefit from understanding task scenes, which inherently entail heterogeneous components such as current objects, goal objects, and environmental constraints. The semantic relationships among these components are distinct from each other and crucial for multi-skilled robots to perform efficiently in everyday scenarios. We propose a hierarchical robotic manipulation system that learns the underlying relationships and maximizes the collaborative power of its diverse skills (e.g., pick-place, push) for rearranging adversarial objects in constrained environments. The high-level coordinator employs a heterogeneous graph neural network (HetGNN), which reasons about the current objects, goal objects, and environmental constraints; the low-level 3D Convolutional Neural Network-based actors execute the action primitives. Our approach is trained entirely in simulation, and achieved an average success rate of 87.88% and a planning cost of 12.82 in real-world experiments, surpassing all baseline methods. Supplementary material is available at https://sites.google.com/umn.edu/versatile-rearrangement.
Intelligent trading strategy based on improved directional change and regime change detection
Authors: Bing Wu, Xiangzu Han
Subjects: Computational Engineering, Finance, and Science (cs.CE)
Abstract
Previous research primarily characterized price movements according to time intervals, resulting in temporal discontinuity and overlooking crucial activities in financial markets. Directional Change (DC) is an alternative approach to sampling price data, highlighting significant points while blurring out noise details in price movements. However, traditional DC treated the thresholds of upward and downward trends with distinct intrinsic patterns as equivalent and preset them as fixed values, which are dependent on the subjective judgment of traders. To enhance the generalization performance of this methodology, we improved DC by introducing a modified threshold selection technique. Specifically, we addressed upward and downward trends distinctly by incorporating a decay coefficient. Further, we simultaneously optimized the threshold and decay coefficient using the Bayesian Optimization Algorithm (BOA). Additionally, we recognized the abnormal market state by regime change detection based on the Hidden Markov Model (RCD-HMM) to reduce the risk. Our Intelligent Trading Algorithm (ITA) was constructed based on above methods and the experiments were carried out on tick data from diverse currency pairs in the forex market. The experimental results showed a significant increase in profit and reduction in risk of DC-based trading strategies, which demonstrated the effectiveness of our proposed methods.
MINS: Efficient and Robust Multisensor-aided Inertial Navigation System
Authors: Woosik Lee, Patrick Geneva, Chuchu Chen, Guoquan Huang
Abstract
Robust multisensor fusion of multi-modal measurements such as IMUs, wheel encoders, cameras, LiDARs, and GPS holds great potential due to its innate ability to improve resilience to sensor failures and measurement outliers, thereby enabling robust autonomy. To the best of our knowledge, this work is among the first to develop a consistent tightly-coupled Multisensor-aided Inertial Navigation System (MINS) that is capable of fusing the most common navigation sensors in an efficient filtering framework, by addressing the particular challenges of computational complexity, sensor asynchronicity, and intra-sensor calibration. In particular, we propose a consistent high-order on-manifold interpolation scheme to enable efficient asynchronous sensor fusion and state management strategy (i.e. dynamic cloning). The proposed dynamic cloning leverages motion-induced information to adaptively select interpolation orders to control computational complexity while minimizing trajectory representation errors. We perform online intrinsic and extrinsic (spatiotemporal) calibration of all onboard sensors to compensate for poor prior calibration and/or degraded calibration varying over time. Additionally, we develop an initialization method with only proprioceptive measurements of IMU and wheel encoders, instead of exteroceptive sensors, which is shown to be less affected by the environment and more robust in highly dynamic scenarios. We extensively validate the proposed MINS in simulations and large-scale challenging real-world datasets, outperforming the existing state-of-the-art methods, in terms of localization accuracy, consistency, and computation efficiency. We have also open-sourced our algorithm, simulator, and evaluation toolbox for the benefit of the community: https://github.com/rpng/mins.
KDD-LOAM: Jointly Learned Keypoint Detector and Descriptors Assisted LiDAR Odometry and Mapping
Authors: Renlang Huang, Minglei Zhao, Jiming Chen, Liang Li
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Sparse keypoint matching based on distinct 3D feature representations can improve the efficiency and robustness of point cloud registration. Existing learning-based 3D descriptors and keypoint detectors are either independent or loosely coupled, so they cannot fully adapt to each other. In this work, we propose a tightly coupled keypoint detector and descriptor (TCKDD) based on a multi-task fully convolutional network with a probabilistic detection loss. In particular, this self-supervised detection loss fully adapts the keypoint detector to any jointly learned descriptors and benefits the self-supervised learning of descriptors. Extensive experiments on both indoor and outdoor datasets show that our TCKDD achieves state-of-the-art performance in point cloud registration. Furthermore, we design a keypoint detector and descriptors-assisted LiDAR odometry and mapping framework (KDD-LOAM), whose real-time odometry relies on keypoint descriptor matching-based RANSAC. The sparse keypoints are further used for efficient scan-to-map registration and mapping. Experiments on KITTI dataset demonstrate that KDD-LOAM significantly surpasses LOAM and shows competitive performance in odometry.
SOCI^+: An Enhanced Toolkit for Secure OutsourcedComputation on Integers
Authors: Bowen Zhao, Weiquan Deng, Xiaoguo Li, Ximeng Liu, Qingqi Pei, Robert H. Deng
Abstract
Secure outsourced computation is critical for cloud computing to safeguard data confidentiality and ensure data usability. Recently, secure outsourced computation schemes following a twin-server architecture based on partially homomorphic cryptosystems have received increasing attention. The Secure Outsourced Computation on Integers (SOCI) [1] toolkit is the state-of-the-art among these schemes which can perform secure computation on integers without requiring the costly bootstrapping operation as in fully homomorphic encryption; however, SOCI suffers from relatively large computation and communication overhead. In this paper, we propose SOCI+ which significantly improves the performance of SOCI. Specifically, SOCI+ employs a novel (2, 2)-threshold Paillier cryptosystem with fast encryption and decryption as its cryptographic primitive, and supports a suite of efficient secure arithmetic computation on integers protocols, including a secure multiplication protocol (SMUL), a secure comparison protocol (SCMP), a secure sign bit-acquisition protocol (SSBA), and a secure division protocol (SDIV), all based on the (2, 2)-threshold Paillier cryptosystem with fast encryption and decryption. In addition, SOCI+ incorporates an offline and online computation mechanism to further optimize its performance. We perform rigorous theoretical analysis to prove the correctness and security of SOCI+. Compared with SOCI, our experimental evaluation shows that SOCI+ is up to 5.4 times more efficient in computation and 40% less in communication overhead.
Prosumers Participation in Markets: A Scalar-Parameterized Function Bidding Approach
Authors: Abdullah Alawad, Muhammad Aneeq uz Zaman, Khaled Alshehri, Tamer Başar
Subjects: Computer Science and Game Theory (cs.GT); Systems and Control (eess.SY)
Abstract
In uniform-price markets, suppliers compete to supply a resource to consumers, resulting in a single market price determined by their competition. For sufficient flexibility, producers and consumers prefer to commit to a function as their strategies, indicating their preferred quantity at any given market price. Producers and consumers may wish to act as both, i.e., prosumers. In this paper, we examine the behavior of profit-maximizing prosumers in a uniform-price market for resource allocation with the objective of maximizing the social welfare. We propose a scalar-parameterized function bidding mechanism for the prosumers, in which we establish the existence and uniqueness of Nash equilibrium. Furthermore, we provide an efficient way to compute the Nash equilibrium through the computation of the market allocation at the Nash equilibrium. Finally, we present a case study to illustrate the welfare loss under different variations of market parameters, such as the market's supply capacity and inelastic demand.
Cardinality Estimation of Subgraph Matching: A Filtering-Sampling Approach
Authors: Wonseok Shin, Siwoo Song, Kunsoo Park, Wook-Shin Han
Abstract
Subgraph counting is a fundamental problem in understanding and analyzing graph structured data, yet computationally challenging. This calls for an accurate and efficient algorithm for Subgraph Cardinality Estimation, which is to estimate the number of all isomorphic embeddings of a query graph in a data graph. We present FaSTest, a novel algorithm that combines (1) a powerful filtering technique to significantly reduce the sample space, (2) an adaptive tree sampling algorithm for accurate and efficient estimation, and (3) a worst-case optimal stratified graph sampling algorithm for difficult instances. Extensive experiments on real-world datasets show that FaSTest outperforms state-of-the-art sampling-based methods by up to two orders of magnitude and GNN-based methods by up to three orders of magnitude in terms of accuracy.
Abstract
With the proliferation of video data in smart city applications like intelligent transportation, efficient video analytics has become crucial but also challenging. This paper proposes a semantics-driven cloud-edge collaborative approach for accelerating video inference, using license plate recognition as a case study. The method separates semantics extraction and recognition, allowing edge servers to only extract visual semantics (license plate patches) from video frames and offload computation-intensive recognition to the cloud or neighboring edges based on load. This segmented processing coupled with a load-aware work distribution strategy aims to reduce end-to-end latency and improve throughput. Experiments demonstrate significant improvements in end-to-end inference speed (up to 5x faster), throughput (up to 9 FPS), and reduced traffic volumes (50% less) compared to cloud-only or edge-only processing, validating the efficiency of the proposed approach. The cloud-edge collaborative framework with semantics-driven work partitioning provides a promising solution for scaling video analytics in smart cities.
In-Hand Re-grasp Manipulation with Passive Dynamic Actions via Imitation Learning
Abstract
Re-grasp manipulation leverages on ergonomic tools to assist humans in accomplishing diverse tasks. In certain scenarios, humans often employ external forces to effortlessly and precisely re-grasp tools like a hammer. Previous development on controllers for in-grasp sliding motion using passive dynamic actions (e.g.,gravity) relies on apprehension of finger-object contact information, and requires customized design for individual objects with varied geometry and weight distribution. It limits their adaptability to diverse objects. In this paper, we propose an end-to-end sliding motion controller based on imitation learning (IL) that necessitates minimal prior knowledge of object mechanics, relying solely on object position information. To expedite training convergence, we utilize a data glove to collect expert data trajectories and train the policy through Generative Adversarial Imitation Learning (GAIL). Simulation results demonstrate the controller's versatility in performing in-hand sliding tasks with objects of varying friction coefficients, geometric shapes, and masses. By migrating to a physical system using visual position estimation, the controller demonstrated an average success rate of 86%, surpassing the baseline algorithm's success rate of 35% of Behavior Cloning(BC) and 20% of Proximal Policy Optimization (PPO).
LogicMP: A Neuro-symbolic Approach for Encoding First-order Logic Constraints
Abstract
Integrating first-order logic constraints (FOLCs) with neural networks is a crucial but challenging problem since it involves modeling intricate correlations to satisfy the constraints. This paper proposes a novel neural layer, LogicMP, whose layers perform mean-field variational inference over an MLN. It can be plugged into any off-the-shelf neural network to encode FOLCs while retaining modularity and efficiency. By exploiting the structure and symmetries in MLNs, we theoretically demonstrate that our well-designed, efficient mean-field iterations effectively mitigate the difficulty of MLN inference, reducing the inference from sequential calculation to a series of parallel tensor operations. Empirical results in three kinds of tasks over graphs, images, and text show that LogicMP outperforms advanced competitors in both performance and efficiency.
Enabling Resource-efficient AIoT System with Cross-level Optimization: A survey
Abstract
The emerging field of artificial intelligence of things (AIoT, AI+IoT) is driven by the widespread use of intelligent infrastructures and the impressive success of deep learning (DL). With the deployment of DL on various intelligent infrastructures featuring rich sensors and weak DL computing capabilities, a diverse range of AIoT applications has become possible. However, DL models are notoriously resource-intensive. Existing research strives to realize near-/realtime inference of AIoT live data and low-cost training using AIoT datasets on resource-scare infrastructures. Accordingly, the accuracy and responsiveness of DL models are bounded by resource availability. To this end, the algorithm-system co-design that jointly optimizes the resource-friendly DL models and model-adaptive system scheduling improves the runtime resource availability and thus pushes the performance boundary set by the standalone level. Unlike previous surveys on resource-friendly DL models or hand-crafted DL compilers/frameworks with partially fine-tuned components, this survey aims to provide a broader optimization space for more free resource-performance tradeoffs. The cross-level optimization landscape involves various granularity, including the DL model, computation graph, operator, memory schedule, and hardware instructor in both on-device and distributed paradigms. Furthermore, due to the dynamic nature of AIoT context, which includes heterogeneous hardware, agnostic sensing data, varying user-specified performance demands, and resource constraints, this survey explores the context-aware inter-/intra-device controllers for automatic cross-level adaptation. Additionally, we identify some potential directions for resource-efficient AIoT systems. By consolidating problems and techniques scattered over diverse levels, we aim to help readers understand their connections and stimulate further discussions.
Patterns Induce Injectivity: A New Thinking in Constructing Injective Local Rules of 1D Cellular Automata over $\mathbb{F}_2$
Abstract
We discovered that certain patterns called injective patterns remain stable during the revolution process, allowing us to create many reversible CA simply by using them to design the revolution rules. By examining injective patterns, we investigated their structural stability during revolutions. This led us to discover extended patterns and pattern mixtures that can create more reversible cellular automata. Furthermore, our research proposed a new way to study the reversibility of CA by observing the structure of local rule $f$. In this paper, we will explicate our study and propose an efficient method for finding the injective patterns. Our algorithms can find injective rules and generate local rule $f$ by traversing $2^{N}$, instead of $2^{2^{N}}$ to check all injective rules and pick the injective ones.
Fast Locality Sensitive Hashing with Theoretical Guarantee
Authors: Zongyuan Tan, Hongya Wang, Bo Xu, Minjie Luo, Ming Du
Subjects: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS)
Abstract
Locality-sensitive hashing (LSH) is an effective randomized technique widely used in many machine learning tasks. The cost of hashing is proportional to data dimensions, and thus often the performance bottleneck when dimensionality is high and the number of hash functions involved is large. Surprisingly, however, little work has been done to improve the efficiency of LSH computation. In this paper, we design a simple yet efficient LSH scheme, named FastLSH, under l2 norm. By combining random sampling and random projection, FastLSH reduces the time complexity from O(n) to O(m) (m<n), where n is the data dimensionality and m is the number of sampled dimensions. Moreover, FastLSH has provable LSH property, which distinguishes it from the non-LSH fast sketches. We conduct comprehensive experiments over a collection of real and synthetic datasets for the nearest neighbor search task. Experimental results demonstrate that FastLSH is on par with the state-of-the-arts in terms of answer quality, space occupation and query efficiency, while enjoying up to 80x speedup in hash function evaluation. We believe that FastLSH is a promising alternative to the classic LSH scheme.
Energy-Efficient Precoding Designs for Multi-User Visible Light Communication Systems with Confidential Messages
Authors: Son T. Duong, Thanh V. Pham, Chuyen T. Nguyen, Anh T. Pham
Subjects: Information Theory (cs.IT); Systems and Control (eess.SY)
Abstract
This paper studies energy-efficient precoding designs for multi-user visible light communication (VLC) systems from the perspective of physical layer security where users' messages must be kept mutually confidential. For such systems, we first derive a lower bound on the achievable secrecy rate of each user. Next, the total power consumption for illumination and data transmission is thoroughly analyzed. We then tackle the problem of maximizing energy efficiency, given that each user's secrecy rate satisfies a certain threshold. The design problem is shown to be non-convex fractional programming, which renders finding the optimal solution computationally prohibitive. Our aim in this paper is, therefore, to find sub-optimal yet low complexity solutions. For this purpose, the traditional Dinkelbach algorithm is first employed to reformulate the original problem to a non-fractional parameterized one. Two different approaches based on the convex-concave procedure (CCCP) and Semidefinite Relaxation (SDR) are utilized to solve the non-convex parameterized problem. In addition, to further reduce the complexity, we investigate a design using the zero-forcing (ZF) technique. Numerical results are conducted to show the feasibility, convergence, and performance of the proposed algorithms depending on different parameters of the system.
VideoAdviser: Video Knowledge Distillation for Multimodal Transfer Learning
Abstract
Multimodal transfer learning aims to transform pretrained representations of diverse modalities into a common domain space for effective multimodal fusion. However, conventional systems are typically built on the assumption that all modalities exist, and the lack of modalities always leads to poor inference performance. Furthermore, extracting pretrained embeddings for all modalities is computationally inefficient for inference. In this work, to achieve high efficiency-performance multimodal transfer learning, we propose VideoAdviser, a video knowledge distillation method to transfer multimodal knowledge of video-enhanced prompts from a multimodal fundamental model (teacher) to a specific modal fundamental model (student). With an intuition that the best learning performance comes with professional advisers and smart students, we use a CLIP-based teacher model to provide expressive multimodal knowledge supervision signals to a RoBERTa-based student model via optimizing a step-distillation objective loss -- first step: the teacher distills multimodal knowledge of video-enhanced prompts from classification logits to a regression logit -- second step: the multimodal knowledge is distilled from the regression logit of the teacher to the student. We evaluate our method in two challenging multimodal tasks: video-level sentiment analysis (MOSI and MOSEI datasets) and audio-visual retrieval (VEGAS dataset). The student (requiring only the text modality as input) achieves an MAE score improvement of up to 12.3% for MOSI and MOSEI. Our method further enhances the state-of-the-art method by 3.4% mAP score for VEGAS without additional computations for inference. These results suggest the strengths of our method for achieving high efficiency-performance multimodal transfer learning.
AdaEvo: Edge-Assisted Continuous and Timely DNN Model Evolution for Mobile Devices
Authors: Lehao Wang, Zhiwen Yu, Haoyi Yu, Sicong Liu, Yaxiong Xie, Bin Guo, Yunxin Liu
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC)
Abstract
Mobile video applications today have attracted significant attention. Deep learning model (e.g. deep neural network, DNN) compression is widely used to enable on-device inference for facilitating robust and private mobile video applications. The compressed DNN, however, is vulnerable to the agnostic data drift of the live video captured from the dynamically changing mobile scenarios. To combat the data drift, mobile ends rely on edge servers to continuously evolve and re-compress the DNN with freshly collected data. We design a framework, AdaEvo, that efficiently supports the resource-limited edge server handling mobile DNN evolution tasks from multiple mobile ends. The key goal of AdaEvo is to maximize the average quality of experience (QoE), e.g. the proportion of high-quality DNN service time to the entire life cycle, for all mobile ends. Specifically, it estimates the DNN accuracy drops at the mobile end without labels and performs a dedicated video frame sampling strategy to control the size of retraining data. In addition, it balances the limited computing and memory resources on the edge server and the competition between asynchronous tasks initiated by different mobile users. With an extensive evaluation of real-world videos from mobile scenarios and across four diverse mobile tasks, experimental results show that AdaEvo enables up to 34% accuracy improvement and 32% average QoE improvement.
Overcoming the Fear of the Dark: Occlusion-Aware Model-Predictive Planning for Automated Vehicles Using Risk Fields
Authors: Chris van der Ploeg, Truls Nyberg, José Manuel Gaspar Sánchez, Emilia Silvas, Nathan van de Wouw
Subjects: Robotics (cs.RO); Optimization and Control (math.OC)
Abstract
As vehicle automation advances, motion planning algorithms face escalating challenges in achieving safe and efficient navigation. Existing Advanced Driver Assistance Systems (ADAS) primarily focus on basic tasks, leaving unexpected scenarios for human intervention, which can be error-prone. Motion planning approaches for higher levels of automation in the state-of-the-art are primarily oriented toward the use of risk- or anti-collision constraints, using over-approximates of the shapes and sizes of other road users to prevent collisions. These methods however suffer from conservative behavior and the risk of infeasibility in high-risk initial conditions. In contrast, our work introduces a novel multi-objective trajectory generation approach. We propose an innovative method for constructing risk fields that accommodates diverse entity shapes and sizes, which allows us to also account for the presence of potentially occluded objects. This methodology is integrated into an occlusion-aware trajectory generator, enabling dynamic and safe maneuvering through intricate environments while anticipating (potentially hidden) road users and traveling along the infrastructure toward a specific goal. Through theoretical underpinnings and simulations, we validate the effectiveness of our approach. This paper bridges crucial gaps in motion planning for automated vehicles, offering a pathway toward safer and more adaptable autonomous navigation in complex urban contexts.
Abstract
Various works have been extensively studied in the research of text-to-image generation. Although existing models perform well in text-to-image generation, there are significant challenges when directly employing them to generate images in dialogs. In this paper, we first highlight a new problem: dialog-to-image generation, that is, given the dialog context, the model should generate a realistic image which is consistent with the specified conversation as response. To tackle the problem, we propose an efficient approach for dialog-to-image generation without any intermediate translation, which maximizes the extraction of the semantic information contained in the dialog. Considering the characteristics of dialog structure, we put segment token before each sentence in a turn of a dialog to differentiate different speakers. Then, we fine-tune pre-trained text-to-image models to enable them to generate images conditioning on processed dialog context. After fine-tuning, our approach can consistently improve the performance of various models across multiple metrics. Experimental results on public benchmark demonstrate the effectiveness and practicability of our method.
Residual Scheduling: A New Reinforcement Learning Approach to Solving Job Shop Scheduling Problem
Abstract
Job-shop scheduling problem (JSP) is a mathematical optimization problem widely used in industries like manufacturing, and flexible JSP (FJSP) is also a common variant. Since they are NP-hard, it is intractable to find the optimal solution for all cases within reasonable times. Thus, it becomes important to develop efficient heuristics to solve JSP/FJSP. A kind of method of solving scheduling problems is construction heuristics, which constructs scheduling solutions via heuristics. Recently, many methods for construction heuristics leverage deep reinforcement learning (DRL) with graph neural networks (GNN). In this paper, we propose a new approach, named residual scheduling, to solving JSP/FJSP. In this new approach, we remove irrelevant machines and jobs such as those finished, such that the states include the remaining (or relevant) machines and jobs only. Our experiments show that our approach reaches state-of-the-art (SOTA) among all known construction heuristics on most well-known open JSP and FJSP benchmarks. In addition, we also observe that even though our model is trained for scheduling problems of smaller sizes, our method still performs well for scheduling problems of large sizes. Interestingly in our experiments, our approach even reaches zero gap for 49 among 50 JSP instances whose job numbers are more than 150 on 20 machines.
Raijū: Reinforcement Learning-Guided Post-Exploitation for Automating Security Assessment of Network Systems
Authors: Van-Hau Pham, Hien Do Hoang, Phan Thanh Trung, Van Dinh Quoc, Trong-Nghia To, Phan The Duy
Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
Abstract
In order to assess the risks of a network system, it is important to investigate the behaviors of attackers after successful exploitation, which is called post-exploitation. Although there are various efficient tools supporting post-exploitation implementation, no application can automate this process. Most of the steps of this process are completed by experts who have profound knowledge of security, known as penetration testers or pen-testers. To this end, our study proposes the Raij\=u framework, a Reinforcement Learning (RL)-driven automation approach that assists pen-testers in quickly implementing the process of post-exploitation for security-level evaluation in network systems. We implement two RL algorithms, Advantage Actor-Critic (A2C) and Proximal Policy Optimization (PPO), to train specialized agents capable of making intelligent actions, which are Metasploit modules to automatically launch attacks of privileges escalation, gathering hashdump, and lateral movement. By leveraging RL, we aim to empower these agents with the ability to autonomously select and execute actions that can exploit vulnerabilities in target systems. This approach allows us to automate certain aspects of the penetration testing workflow, making it more efficient and responsive to emerging threats and vulnerabilities. The experiments are performed in four real environments with agents trained in thousands of episodes. The agents automatically select actions and launch attacks on the environments and achieve over 84\% of successful attacks with under 55 attack steps given. Moreover, the A2C algorithm has proved extremely effective in the selection of proper actions for automation of post-exploitation.
Defending Against Physical Adversarial Patch Attacks on Infrared Human Detection
Abstract
Infrared detection is an emerging technique for safety-critical tasks owing to its remarkable anti-interference capability. However, recent studies have revealed that it is vulnerable to physically-realizable adversarial patches, posing risks in its real-world applications. To address this problem, we are the first to investigate defense strategies against adversarial patch attacks on infrared detection, especially human detection. We have devised a straightforward defense strategy, patch-based occlusion-aware detection (POD), which efficiently augments training samples with random patches and subsequently detects them. POD not only robustly detects people but also identifies adversarial patch locations. Surprisingly, while being extremely computationally efficient, POD easily generalizes to state-of-the-art adversarial patch attacks that are unseen during training. Furthermore, POD improves detection precision even in a clean (i.e., no-patch) situation due to the data augmentation effect. Evaluation demonstrated that POD is robust to adversarial patches of various shapes and sizes. The effectiveness of our baseline approach is shown to be a viable defense mechanism for real-world infrared human detection systems, paving the way for exploring future research directions.
SAF-Net: Self-Attention Fusion Network for Myocardial Infarction Detection using Multi-View Echocardiography
Abstract
Myocardial infarction (MI) is a severe case of coronary artery disease (CAD) and ultimately, its detection is substantial to prevent progressive damage to the myocardium. In this study, we propose a novel view-fusion model named self-attention fusion network (SAF-Net) to detect MI from multi-view echocardiography recordings. The proposed framework utilizes apical 2-chamber (A2C) and apical 4-chamber (A4C) view echocardiography recordings for classification. Three reference frames are extracted from each recording of both views and deployed pre-trained deep networks to extract highly representative features. The SAF-Net model utilizes a self-attention mechanism to learn dependencies in extracted feature vectors. The proposed model is computationally efficient thanks to its compact architecture having three main parts: a feature embedding to reduce dimensionality, self-attention for view-pooling, and dense layers for the classification. Experimental evaluation is performed using the HMC-QU-TAU dataset which consists of 160 patients with A2C and A4C view echocardiography recordings. The proposed SAF-Net model achieves a high-performance level with 88.26% precision, 77.64% sensitivity, and 78.13% accuracy. The results demonstrate that the SAF-Net model achieves the most accurate MI detection over multi-view echocardiography recordings.
Improving Facade Parsing with Vision Transformers and Line Integration
Authors: Bowen Wang, Jiaxing Zhang, Ran Zhang, Yunqin Li, Liangzhi Li, Yuta Nakashima
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Facade parsing stands as a pivotal computer vision task with far-reaching applications in areas like architecture, urban planning, and energy efficiency. Despite the recent success of deep learning-based methods in yielding impressive results on certain open-source datasets, their viability for real-world applications remains uncertain. Real-world scenarios are considerably more intricate, demanding greater computational efficiency. Existing datasets often fall short in representing these settings, and previous methods frequently rely on extra models to enhance accuracy, which requires much computation cost. In this paper, we introduce Comprehensive Facade Parsing (CFP), a dataset meticulously designed to encompass the intricacies of real-world facade parsing tasks. Comprising a total of 602 high-resolution street-view images, this dataset captures a diverse array of challenging scenarios, including sloping angles and densely clustered buildings, with painstakingly curated annotations for each image. We introduce a new pipeline known as Revision-based Transformer Facade Parsing (RTFP). This marks the pioneering utilization of Vision Transformers (ViT) in facade parsing, and our experimental results definitively substantiate its merit. We also design Line Acquisition, Filtering, and Revision (LAFR), an efficient yet accurate revision algorithm that can improve the segment result solely from simple line detection using prior knowledge of the facade. In ECP 2011, RueMonge 2014, and our CFP, we evaluate the superiority of our method. The dataset and code are available at https://github.com/wbw520/RTFP.
Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models
Authors: Jung Hwan Heo, Jeonghoon Kim, Beomseok Kwon, Byeongwook Kim, Se Jung Kwon, Dongsoo Lee
Abstract
Large Language Models (LLMs) have recently demonstrated a remarkable success across various tasks. However, efficiently serving LLMs has been a challenge due to its large memory bottleneck, specifically in small batch inference settings (e.g. mobile devices). Weight-only quantization can be a promising approach, but sub-4 bit quantization remains a challenge due to large-magnitude activation outliers. To mitigate the undesirable outlier effect, we first propose per-IC quantization, a simple yet effective method that creates quantization groups within each input channel (IC) rather than the conventional per-output channel (OC). Our method is motivated by the observation that activation outliers affect the input dimension of the weight matrix, so similarly grouping the weights in the IC direction can isolate outliers to be within a group. We also find that activation outliers do not dictate quantization difficulty, and inherent weight sensitivities also exist. With per-IC quantization as a new outlier-friendly scheme, we then propose Adaptive Dimensions (AdaDim), a versatile quantization framework that can adapt to various weight sensitivity patterns. We demonstrate the effectiveness of AdaDim by augmenting prior methods such as Round-To-Nearest and GPTQ, showing significant improvements across various language modeling benchmarks for both base (up to +4.7% on MMLU) and instruction-tuned (up to +10% on HumanEval) LLMs.
Towards Faithful Neural Network Intrinsic Interpretation with Shapley Additive Self-Attribution
Abstract
Self-interpreting neural networks have garnered significant interest in research. Existing works in this domain often (1) lack a solid theoretical foundation ensuring genuine interpretability or (2) compromise model expressiveness. In response, we formulate a generic Additive Self-Attribution (ASA) framework. Observing the absence of Shapley value in Additive Self-Attribution, we propose Shapley Additive Self-Attributing Neural Network (SASANet), with theoretical guarantees for the self-attribution value equal to the output's Shapley values. Specifically, SASANet uses a marginal contribution-based sequential schema and internal distillation-based training strategies to model meaningful outputs for any number of features, resulting in un-approximated meaningful value function. Our experimental results indicate SASANet surpasses existing self-attributing models in performance and rivals black-box models. Moreover, SASANet is shown more precise and efficient than post-hoc methods in interpreting its own predictions.
Jointly Training Large Autoregressive Multimodal Models
Authors: Emanuele Aiello, Lili Yu, Yixin Nie, Armen Aghajanyan, Barlas Oguz
Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
Abstract
In recent years, advances in the large-scale pretraining of language and text-to-image models have revolutionized the field of machine learning. Yet, integrating these two modalities into a single, robust model capable of generating seamless multimodal outputs remains a significant challenge. To address this gap, we present the Joint Autoregressive Mixture (JAM) framework, a modular approach that systematically fuses existing text and image generation models. We also introduce a specialized, data-efficient instruction-tuning strategy, tailored for mixed-modal generation tasks. Our final instruct-tuned model demonstrates unparalleled performance in generating high-quality multimodal outputs and represents the first model explicitly designed for this purpose.
HPL-ViT: A Unified Perception Framework for Heterogeneous Parallel LiDARs in V2V
Abstract
To develop the next generation of intelligent LiDARs, we propose a novel framework of parallel LiDARs and construct a hardware prototype in our experimental platform, DAWN (Digital Artificial World for Natural). It emphasizes the tight integration of physical and digital space in LiDAR systems, with networking being one of its supported core features. In the context of autonomous driving, V2V (Vehicle-to-Vehicle) technology enables efficient information sharing between different agents which significantly promotes the development of LiDAR networks. However, current research operates under an ideal situation where all vehicles are equipped with identical LiDAR, ignoring the diversity of LiDAR categories and operating frequencies. In this paper, we first utilize OpenCDA and RLS (Realistic LiDAR Simulation) to construct a novel heterogeneous LiDAR dataset named OPV2V-HPL. Additionally, we present HPL-ViT, a pioneering architecture designed for robust feature fusion in heterogeneous and dynamic scenarios. It uses a graph-attention Transformer to extract domain-specific features for each agent, coupled with a cross-attention mechanism for the final fusion. Extensive experiments on OPV2V-HPL demonstrate that HPL-ViT achieves SOTA (state-of-the-art) performance in all settings and exhibits outstanding generalization capabilities.
Advancing the distributed Multi-GPU ChASE library through algorithm optimization and NCCL library
Authors: Xinzhe Wu (1), Edoardo Di Napoli (1) ((1) Juelich Supercomputing Centre, Forschungszentrum Juelich, Germany)
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC); Computational Engineering, Finance, and Science (cs.CE)
Abstract
As supercomputers become larger with powerful Graphics Processing Unit (GPU), traditional direct eigensolvers struggle to keep up with the hardware evolution and scale efficiently due to communication and synchronization demands. Conversely, subspace eigensolvers, like the Chebyshev Accelerated Subspace Eigensolver (ChASE), have a simpler structure and can overcome communication and synchronization bottlenecks. ChASE is a modern subspace eigensolver that uses Chebyshev polynomials to accelerate the computation of extremal eigenpairs of dense Hermitian eigenproblems. In this work we show how we have modified ChASE by rethinking its memory layout, introducing a novel parallelization scheme, switching to a more performing communication-avoiding algorithm for one of its inner modules, and substituting the MPI library by the vendor-optimized NCCL library. The resulting library can tackle dense problems with size up to $N=\mathcal{O}(10^6)$, and scales effortlessly up to the full 900 nodes -- each one powered by 4$\times$A100 NVIDIA GPUs -- of the JUWELS Booster hosted at the J\"ulich Supercomputing Centre.
PolarNet: 3D Point Clouds for Language-Guided Robotic Manipulation
Authors: Shizhe Chen, Ricardo Garcia, Cordelia Schmid, Ivan Laptev
Subjects: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
Abstract
The ability for robots to comprehend and execute manipulation tasks based on natural language instructions is a long-term goal in robotics. The dominant approaches for language-guided manipulation use 2D image representations, which face difficulties in combining multi-view cameras and inferring precise 3D positions and relationships. To address these limitations, we propose a 3D point cloud based policy called PolarNet for language-guided manipulation. It leverages carefully designed point cloud inputs, efficient point cloud encoders, and multimodal transformers to learn 3D point cloud representations and integrate them with language instructions for action prediction. PolarNet is shown to be effective and data efficient in a variety of experiments conducted on the RLBench benchmark. It outperforms state-of-the-art 2D and 3D approaches in both single-task and multi-task learning. It also achieves promising results on a real robot.
RapidEarth: A Search-by-Classification Engine for Large-Scale Geospatial Imagery
Authors: Christian Lülf, Denis Mayr Lima Martins, Marcos Antonio Vaz Salles, Yongluan Zhou, Fabian Gieseke
Abstract
Data exploration and analysis in various domains often necessitate the search for specific objects in massive databases. A common search strategy, often known as search-by-classification, resorts to training machine learning models on small sets of positive and negative samples and to performing inference on the entire database to discover additional objects of interest. While such an approach often yields very good results in terms of classification performance, the entire database usually needs to be scanned, a process that can easily take several hours even for medium-sized data catalogs. In this work, we present RapidEarth, a geospatial search-by-classification engine that allows analysts to rapidly search for interesting objects in very large data collections of satellite imagery in a matter of seconds, without the need to scan the entire data catalog. RapidEarth embodies a co-design of multidimensional indexing structures and decision branches, a recently proposed variant of classical decision trees. These decision branches allow RapidEarth to transform the inference phase into a set of range queries, which can be efficiently processed by leveraging the aforementioned multidimensional indexing structures. The main contribution of this work is a geospatial search engine that implements these technical findings.
Design and Optimization of Residual Neural Network Accelerators for Low-Power FPGAs Using High-Level Synthesis
Authors: Filippo Minnella, Teodoro Urso, Mihai T. Lazarescu, Luciano Lavagno
Subjects: Hardware Architecture (cs.AR); Signal Processing (eess.SP)
Abstract
Residual neural networks are widely used in computer vision tasks. They enable the construction of deeper and more accurate models by mitigating the vanishing gradient problem. Their main innovation is the residual block which allows the output of one layer to bypass one or more intermediate layers and be added to the output of a later layer. Their complex structure and the buffering required by the residual block make them difficult to implement on resource-constrained platforms. We present a novel design flow for implementing deep learning models for field programmable gate arrays optimized for ResNets, using a strategy to reduce their buffering overhead to obtain a resource-efficient implementation of the residual layer. Our high-level synthesis (HLS)-based flow encompasses a thorough set of design principles and optimization strategies, exploiting in novel ways standard techniques such as temporal reuse and loop merging to efficiently map ResNet models, and potentially other skip connection-based NN architectures, into FPGA. The models are quantized to 8-bit integers for both weights and activations, 16-bit for biases, and 32-bit for accumulations. The experimental results are obtained on the CIFAR-10 dataset using ResNet8 and ResNet20 implemented with Xilinx FPGAs using HLS on the Ultra96-V2 and Kria KV260 boards. Compared to the state-of-the-art on the Kria KV260 board, our ResNet20 implementation achieves 2.88X speedup with 0.5% higher accuracy of 91.3%, while ResNet8 accuracy improves by 2.8% to 88.7%. The throughputs of ResNet8 and ResNet20 are 12971 FPS and 3254 FPS on the Ultra96 board, and 30153 FPS and 7601 FPS on the Kria KV26, respectively. They Pareto-dominate state-of-the-art solutions concerning accuracy, throughput, and energy.
Abstract
The classic problem of exact subgraph matching returns those subgraphs in a large-scale data graph that are isomorphic to a given query graph, which has gained increasing importance in many real-world applications such as social network analysis, knowledge graph discovery in the Semantic Web, bibliographical network mining, and so on. In this paper, we propose a novel and effective \textit{graph neural network (GNN)-based path embedding framework} (GNN-PE), which allows efficient exact subgraph matching without introducing false dismissals. Unlike traditional GNN-based graph embeddings that only produce approximate subgraph matching results, in this paper, we carefully devise GNN-based embeddings for paths, such that: if two paths (and 1-hop neighbors of vertices on them) have the subgraph relationship, their corresponding GNN-based embedding vectors will strictly follow the dominance relationship. With such a newly designed property of path dominance embeddings, we are able to propose effective pruning strategies based on path label/dominance embeddings and guarantee no false dismissals for subgraph matching. We build multidimensional indexes over path embedding vectors, and develop an efficient subgraph matching algorithm by traversing indexes over graph partitions in parallel and applying our pruning methods. We also propose a cost-model-based query plan that obtains query paths from the query graph with low query cost. Through extensive experiments, we confirm the efficiency and effectiveness of our proposed GNN-PE approach for exact subgraph matching on both real and synthetic graph data.
Tactile-based Active Inference for Force-Controlled Peg-in-Hole Insertions
Authors: Tatsuya Kamijo, Ixchel G. Ramirez-Alpizar, Enrique Coronado, Gentiane Venture
Abstract
Reinforcement Learning (RL) has shown great promise for efficiently learning force control policies in peg-in-hole tasks. However, robots often face difficulties due to visual occlusions by the gripper and uncertainties in the initial grasping pose of the peg. These challenges often restrict force-controlled insertion policies to situations where the peg is rigidly fixed to the end-effector. While vision-based tactile sensors offer rich tactile feedback that could potentially address these issues, utilizing them to learn effective tactile policies is both computationally intensive and difficult to generalize. In this paper, we propose a robust tactile insertion policy that can align the tilted peg with the hole using active inference, without the need for extensive training on large datasets. Our approach employs a dual-policy architecture: one policy focuses on insertion, integrating force control and RL to guide the object into the hole, while the other policy performs active inference based on tactile feedback to align the tilted peg with the hole. In real-world experiments, our dual-policy architecture achieved 90% success rate into a hole with a clearance of less than 0.1 mm, significantly outperforming previous methods that lack tactile sensory feedback (5%). To assess the generalizability of our alignment policy, we conducted experiments with five different pegs, demonstrating its effective adaptation to multiple objects.
Improving Autonomous Driving Safety with POP: A Framework for Accurate Partially Observed Trajectory Predictions
Authors: Sheng Wang, Yingbing Chen, Jie Cheng, Xiaodong Mei, Yongkang Song, Ming Liu
Abstract
Accurate trajectory prediction is crucial for safe and efficient autonomous driving, but handling partial observations presents significant challenges. To address this, we propose a novel trajectory prediction framework called Partial Observations Prediction (POP) for congested urban road scenarios. The framework consists of two stages: self-supervised learning (SSL) and feature distillation. In SSL, a reconstruction branch reconstructs the hidden history of partial observations using a mask procedure and reconstruction head. The feature distillation stage transfers knowledge from a fully observed teacher model to a partially observed student model, improving prediction accuracy. POP achieves comparable results to top-performing methods in open-loop experiments and outperforms the baseline method in closed-loop simulations, including safety metrics. Qualitative results illustrate the superiority of POP in providing reasonable and safe trajectory predictions.
Physics-Based Rigid Body Object Tracking and Friction Filtering From RGB-D Videos
Authors: Rama Krishna Kandukuri, Michael Strecke, Joerg Stueckler
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Physics-based understanding of object interactions from sensory observations is an essential capability in augmented reality and robotics. It enables capturing the properties of a scene for simulation and control. In this paper, we propose a novel approach for real-to-sim which tracks rigid objects in 3D from RGB-D images and infers physical properties of the objects. We use a differentiable physics simulation as state-transition model in an Extended Kalman Filter which can model contact and friction for arbitrary mesh-based shapes and in this way estimate physically plausible trajectories. We demonstrate that our approach can filter position, orientation, velocities, and concurrently can estimate the coefficient of friction of the objects. We analyse our approach on various sliding scenarios in synthetic image sequences of single objects and colliding objects. We also demonstrate and evaluate our approach on a real-world dataset. We will make our novel benchmark datasets publicly available to foster future research in this novel problem setting and comparison with our method.
MindGPT: Interpreting What You See with Non-invasive Brain Recordings
Authors: Jiaxuan Chen, Yu Qi, Yueming Wang, Gang Pan
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Abstract
Decoding of seen visual contents with non-invasive brain recordings has important scientific and practical values. Efforts have been made to recover the seen images from brain signals. However, most existing approaches cannot faithfully reflect the visual contents due to insufficient image quality or semantic mismatches. Compared with reconstructing pixel-level visual images, speaking is a more efficient and effective way to explain visual information. Here we introduce a non-invasive neural decoder, termed as MindGPT, which interprets perceived visual stimuli into natural languages from fMRI signals. Specifically, our model builds upon a visually guided neural encoder with a cross-attention mechanism, which permits us to guide latent neural representations towards a desired language semantic direction in an end-to-end manner by the collaborative use of the large language model GPT. By doing so, we found that the neural representations of the MindGPT are explainable, which can be used to evaluate the contributions of visual properties to language semantics. Our experiments show that the generated word sequences truthfully represented the visual information (with essential details) conveyed in the seen stimuli. The results also suggested that with respect to language decoding tasks, the higher visual cortex (HVC) is more semantically informative than the lower visual cortex (LVC), and using only the HVC can recover most of the semantic information. The code of the MindGPT model will be publicly available at https://github.com/JxuanC/MindGPT.
Generating Transferable Adversarial Simulation Scenarios for Self-Driving via Neural Rendering
Abstract
Self-driving software pipelines include components that are learned from a significant number of training examples, yet it remains challenging to evaluate the overall system's safety and generalization performance. Together with scaling up the real-world deployment of autonomous vehicles, it is of critical importance to automatically find simulation scenarios where the driving policies will fail. We propose a method that efficiently generates adversarial simulation scenarios for autonomous driving by solving an optimal control problem that aims to maximally perturb the policy from its nominal trajectory. Given an image-based driving policy, we show that we can inject new objects in a neural rendering representation of the deployment scene, and optimize their texture in order to generate adversarial sensor inputs to the policy. We demonstrate that adversarial scenarios discovered purely in the neural renderer (surrogate scene) can often be successfully transferred to the deployment scene, without further optimization. We demonstrate this transfer occurs both in simulated and real environments, provided the learned surrogate scene is sufficiently close to the deployment scene.
Learning the Efficient Frontier
Authors: Philippe Chatigny, Ivan Sergienko, Ryan Ferguson, Jordan Weir, Maxime Bergeron
Subjects: Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE)
Abstract
The efficient frontier (EF) is a fundamental resource allocation problem where one has to find an optimal portfolio maximizing a reward at a given level of risk. This optimal solution is traditionally found by solving a convex optimization problem. In this paper, we introduce NeuralEF: a fast neural approximation framework that robustly forecasts the result of the EF convex optimization problem with respect to heterogeneous linear constraints and variable number of optimization inputs. By reformulating an optimization problem as a sequence to sequence problem, we show that NeuralEF is a viable solution to accelerate large-scale simulation while handling discontinuous behavior.
Partial Transport for Point-Cloud Registration
Authors: Yikun Bai, Huy Tran, Steven B. Damelin, Soheil Kolouri
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
Point cloud registration plays a crucial role in various fields, including robotics, computer graphics, and medical imaging. This process involves determining spatial relationships between different sets of points, typically within a 3D space. In real-world scenarios, complexities arise from non-rigid movements and partial visibility, such as occlusions or sensor noise, making non-rigid registration a challenging problem. Classic non-rigid registration methods are often computationally demanding, suffer from unstable performance, and, importantly, have limited theoretical guarantees. The optimal transport problem and its unbalanced variations (e.g., the optimal partial transport problem) have emerged as powerful tools for point-cloud registration, establishing a strong benchmark in this field. These methods view point clouds as empirical measures and provide a mathematically rigorous way to quantify the `correspondence' between (the transformed) source and target points. In this paper, we approach the point-cloud registration problem through the lens of optimal transport theory and first propose a comprehensive set of non-rigid registration methods based on the optimal partial transportation problem. Subsequently, leveraging the emerging work on efficient solutions to the one-dimensional optimal partial transport problem, we extend our proposed algorithms via slicing to gain significant computational efficiency, resulting in fast and robust non-rigid registration algorithms. We demonstrate the effectiveness of our proposed methods and compare them against baselines on various 3D and 2D non-rigid registration problems where the source and target point clouds are corrupted by random noise.
Convolutional Networks with Oriented 1D Kernels
Authors: Alexandre Kirchmeyer, Jia Deng
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
In computer vision, 2D convolution is arguably the most important operation performed by a ConvNet. Unsurprisingly, it has been the focus of intense software and hardware optimization and enjoys highly efficient implementations. In this work, we ask an intriguing question: can we make a ConvNet work without 2D convolutions? Surprisingly, we find that the answer is yes -- we show that a ConvNet consisting entirely of 1D convolutions can do just as well as 2D on ImageNet classification. Specifically, we find that one key ingredient to a high-performing 1D ConvNet is oriented 1D kernels: 1D kernels that are oriented not just horizontally or vertically, but also at other angles. Our experiments show that oriented 1D convolutions can not only replace 2D convolutions but also augment existing architectures with large kernels, leading to improved accuracy with minimal FLOPs increase. A key contribution of this work is a highly-optimized custom CUDA implementation of oriented 1D kernels, specialized to the depthwise convolution setting. Our benchmarks demonstrate that our custom CUDA implementation almost perfectly realizes the theoretical advantage of 1D convolution: it is faster than a native horizontal convolution for any arbitrary angle. Code is available at https://github.com/princeton-vl/Oriented1D.
Show-1: Marrying Pixel and Latent Diffusion Models for Text-to-Video Generation
Authors: David Junhao Zhang, Jay Zhangjie Wu, Jia-Wei Liu, Rui Zhao, Lingmin Ran, Yuchao Gu, Difei Gao, Mike Zheng Shou
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Significant advancements have been achieved in the realm of large-scale pre-trained text-to-video Diffusion Models (VDMs). However, previous methods either rely solely on pixel-based VDMs, which come with high computational costs, or on latent-based VDMs, which often struggle with precise text-video alignment. In this paper, we are the first to propose a hybrid model, dubbed as Show-1, which marries pixel-based and latent-based VDMs for text-to-video generation. Our model first uses pixel-based VDMs to produce a low-resolution video of strong text-video correlation. After that, we propose a novel expert translation method that employs the latent-based VDMs to further upsample the low-resolution video to high resolution. Compared to latent VDMs, Show-1 can produce high-quality videos of precise text-video alignment; Compared to pixel VDMs, Show-1 is much more efficient (GPU memory usage during inference is 15G vs 72G). We also validate our model on standard video generation benchmarks. Our code and model weights are publicly available at \url{https://github.com/showlab/Show-1}.
OrthoPlanes: A Novel Representation for Better 3D-Awareness of GANs
Authors: Honglin He, Zhuoqian Yang, Shikai Li, Bo Dai, Wayne Wu
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Abstract
We present a new method for generating realistic and view-consistent images with fine geometry from 2D image collections. Our method proposes a hybrid explicit-implicit representation called \textbf{OrthoPlanes}, which encodes fine-grained 3D information in feature maps that can be efficiently generated by modifying 2D StyleGANs. Compared to previous representations, our method has better scalability and expressiveness with clear and explicit information. As a result, our method can handle more challenging view-angles and synthesize articulated objects with high spatial degree of freedom. Experiments demonstrate that our method achieves state-of-the-art results on FFHQ and SHHQ datasets, both quantitatively and qualitatively. Project page: \url{https://orthoplanes.github.io/}.
Keyword: faster
RAGAS: Automated Evaluation of Retrieval Augmented Generation
Authors: Shahul Es, Jithin James, Luis Espinosa-Anke, Steven Schockaert
Abstract
We introduce RAGAs (Retrieval Augmented Generation Assessment), a framework for reference-free evaluation of Retrieval Augmented Generation (RAG) pipelines. RAG systems are composed of a retrieval and an LLM based generation module, and provide LLMs with knowledge from a reference textual database, which enables them to act as a natural language layer between a user and textual databases, reducing the risk of hallucinations. Evaluating RAG architectures is, however, challenging because there are several dimensions to consider: the ability of the retrieval system to identify relevant and focused context passages, the ability of the LLM to exploit such passages in a faithful way, or the quality of the generation itself. With RAGAs, we put forward a suite of metrics which can be used to evaluate these different dimensions \textit{without having to rely on ground truth human annotations}. We posit that such a framework can crucially contribute to faster evaluation cycles of RAG architectures, which is especially important given the fast adoption of LLMs.
A Physics Enhanced Residual Learning (PERL) Framework for Traffic State Prediction
Abstract
In vehicle trajectory prediction, physics models and data-driven models are two predominant methodologies. However, each approach presents its own set of challenges: physics models fall short in predictability, while data-driven models lack interpretability. Addressing these identified shortcomings, this paper proposes a novel framework, the Physics-Enhanced Residual Learning (PERL) model. PERL integrates the strengths of physics-based and data-driven methods for traffic state prediction. PERL contains a physics model and a residual learning model. Its prediction is the sum of the physics model result and a predicted residual as a correction to it. It preserves the interpretability inherent to physics-based models and has reduced data requirements compared to data-driven methods. Experiments were conducted using a real-world vehicle trajectory dataset. We proposed a PERL model, with the Intelligent Driver Model (IDM) as its physics car-following model and Long Short-Term Memory (LSTM) as its residual learning model. We compare this PERL model with the physics car-following model, data-driven model, and other physics-informed neural network (PINN) models. The result reveals that PERL achieves better prediction with a small dataset, compared to the physics model, data-driven model, and PINN model. Second, the PERL model showed faster convergence during training, offering comparable performance with fewer training samples than the data-driven model and PINN model. Sensitivity analysis also proves comparable performance of PERL using another residual learning model and a physics car-following model.
MAPTree: Beating "Optimal" Decision Trees with Bayesian Decision Trees
Authors: Colin Sullivan, Mo Tiwari, Sebastian Thrun
Abstract
Decision trees remain one of the most popular machine learning models today, largely due to their out-of-the-box performance and interpretability. In this work, we present a Bayesian approach to decision tree induction via maximum a posteriori inference of a posterior distribution over trees. We first demonstrate a connection between maximum a posteriori inference of decision trees and AND/OR search. Using this connection, we propose an AND/OR search algorithm, dubbed MAPTree, which is able to recover the maximum a posteriori tree. Lastly, we demonstrate the empirical performance of the maximum a posteriori tree both on synthetic data and in real world settings. On 16 real world datasets, MAPTree either outperforms baselines or demonstrates comparable performance but with much smaller trees. On a synthetic dataset, MAPTree also demonstrates greater robustness to noise and better generalization than existing approaches. Finally, MAPTree recovers the maxiumum a posteriori tree faster than existing sampling approaches and, in contrast with those algorithms, is able to provide a certificate of optimality. The code for our experiments is available at https://github.com/ThrunGroup/maptree.
Neural Operators for Accelerating Scientific Simulations and Design
Authors: Kamyar Azzizadenesheli, Nikola Kovachki, Zongyi Li, Miguel Liu-Schiaffini, Jean Kossaifi, Anima Anandkumar
Abstract
Scientific discovery and engineering design are currently limited by the time and cost of physical experiments, selected mostly through trial-and-error and intuition that require deep domain expertise. Numerical simulations present an alternative to physical experiments, but are usually infeasible for complex real-world domains due to the computational requirements of existing numerical methods. Artificial intelligence (AI) presents a potential paradigm shift through the development of fast data-driven surrogate models. In particular, an AI framework, known as neural operators, presents a principled framework for learning mappings between functions defined on continuous domains, e.g., spatiotemporal processes and partial differential equations (PDE). They can extrapolate and predict solutions at new locations unseen during training, i.e., perform zero-shot super-resolution. Neural operators can augment or even replace existing simulators in many applications, such as computational fluid dynamics, weather forecasting, and material modeling, while being 4-5 orders of magnitude faster. Further, neural operators can be integrated with physics and other domain constraints enforced at finer resolutions to obtain high-fidelity solutions and good generalization. Since neural operators are differentiable, they can directly optimize parameters for inverse design and other inverse problems. We believe that neural operators present a transformative approach to simulation and design, enabling rapid research and development.
Abstract
With the proliferation of video data in smart city applications like intelligent transportation, efficient video analytics has become crucial but also challenging. This paper proposes a semantics-driven cloud-edge collaborative approach for accelerating video inference, using license plate recognition as a case study. The method separates semantics extraction and recognition, allowing edge servers to only extract visual semantics (license plate patches) from video frames and offload computation-intensive recognition to the cloud or neighboring edges based on load. This segmented processing coupled with a load-aware work distribution strategy aims to reduce end-to-end latency and improve throughput. Experiments demonstrate significant improvements in end-to-end inference speed (up to 5x faster), throughput (up to 9 FPS), and reduced traffic volumes (50% less) compared to cloud-only or edge-only processing, validating the efficiency of the proposed approach. The cloud-edge collaborative framework with semantics-driven work partitioning provides a promising solution for scaling video analytics in smart cities.
P2I-NET: Mapping Camera Pose to Image via Adversarial Learning for New View Synthesis in Real Indoor Environments
Abstract
Given a new $6DoF$ camera pose in an indoor environment, we study the challenging problem of predicting the view from that pose based on a set of reference RGBD views. Existing explicit or implicit 3D geometry construction methods are computationally expensive while those based on learning have predominantly focused on isolated views of object categories with regular geometric structure. Differing from the traditional \textit{render-inpaint} approach to new view synthesis in the real indoor environment, we propose a conditional generative adversarial neural network (P2I-NET) to directly predict the new view from the given pose. P2I-NET learns the conditional distribution of the images of the environment for establishing the correspondence between the camera pose and its view of the environment, and achieves this through a number of innovative designs in its architecture and training lost function. Two auxiliary discriminator constraints are introduced for enforcing the consistency between the pose of the generated image and that of the corresponding real world image in both the latent feature space and the real world pose space. Additionally a deep convolutional neural network (CNN) is introduced to further reinforce this consistency in the pixel space. We have performed extensive new view synthesis experiments on real indoor datasets. Results show that P2I-NET has superior performance against a number of NeRF based strong baseline models. In particular, we show that P2I-NET is 40 to 100 times faster than these competitor techniques while synthesising similar quality images. Furthermore, we contribute a new publicly available indoor environment dataset containing 22 high resolution RGBD videos where each frame also has accurate camera pose parameters.
Uncertainty Quantification via Neural Posterior Principal Components
Authors: Elias Nehme, Omer Yair, Tomer Michaeli
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV); Machine Learning (stat.ML)
Abstract
Uncertainty quantification is crucial for the deployment of image restoration models in safety-critical domains, like autonomous driving and biological imaging. To date, methods for uncertainty visualization have mainly focused on per-pixel estimates. However, a heatmap of per-pixel variances is typically of little practical use, as it does not capture the strong correlations between pixels. A more natural measure of uncertainty corresponds to the variances along the principal components (PCs) of the posterior distribution. Theoretically, the PCs can be computed by applying PCA on samples generated from a conditional generative model for the input image. However, this requires generating a very large number of samples at test time, which is painfully slow with the current state-of-the-art (diffusion) models. In this work, we present a method for predicting the PCs of the posterior distribution for any input image, in a single forward pass of a neural network. Our method can either wrap around a pre-trained model that was trained to minimize the mean square error (MSE), or can be trained from scratch to output both a predicted image and the posterior PCs. We showcase our method on multiple inverse problems in imaging, including denoising, inpainting, super-resolution, and biological image-to-image translation. Our method reliably conveys instance-adaptive uncertainty directions, achieving uncertainty quantification comparable with posterior samplers while being orders of magnitude faster. Examples are available at https://eliasnehme.github.io/NPPC/
Rapid Network Adaptation: Learning to Adapt Neural Networks Using Test-Time Feedback
Authors: Teresa Yeo, Oğuzhan Fatih Kar, Zahra Sodagar, Amir Zamir
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
We propose a method for adapting neural networks to distribution shifts at test-time. In contrast to training-time robustness mechanisms that attempt to anticipate and counter the shift, we create a closed-loop system and make use of a test-time feedback signal to adapt a network on the fly. We show that this loop can be effectively implemented using a learning-based function, which realizes an amortized optimizer for the network. This leads to an adaptation method, named Rapid Network Adaptation (RNA), that is notably more flexible and orders of magnitude faster than the baselines. Through a broad set of experiments using various adaptation signals and target tasks, we study the efficiency and flexibility of this method. We perform the evaluations using various datasets (Taskonomy, Replica, ScanNet, Hypersim, COCO, ImageNet), tasks (depth, optical flow, semantic segmentation, classification), and distribution shifts (Cross-datasets, 2D and 3D Common Corruptions) with promising results. We end with a discussion on general formulations for handling distribution shifts and our observations from comparing with similar approaches from other domains.
One For All: Video Conversation is Feasible Without Video Instruction Tuning
Authors: Ruyang Liu, Chen Li, Yixiao Ge, Ying Shan, Thomas H. Li, Ge Li
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
The recent progress in Large Language Models (LLM) has spurred various advancements in image-language conversation agents, while how to build a proficient video-based dialogue system is still under exploration. Considering the extensive scale of LLM and visual backbone, minimal GPU memory is left for facilitating effective temporal modeling, which is crucial for comprehending and providing feedback on videos. To this end, we propose Branching Temporal Adapter (BT-Adapter), a novel method for extending image-language pretrained models into the video domain. Specifically, BT-Adapter serves as a plug-and-use temporal modeling branch alongside the pretrained visual encoder, which is tuned while keeping the backbone frozen. Just pretrained once, BT-Adapter can be seamlessly integrated into all image conversation models using this version of CLIP, enabling video conversations without the need for video instructions. Besides, we develop a unique asymmetric token masking strategy inside the branch with tailor-made training tasks for BT-Adapter, facilitating faster convergence and better results. Thanks to BT-Adapter, we are able to empower existing multimodal dialogue models with strong video understanding capabilities without incurring excessive GPU costs. Without bells and whistles, BT-Adapter achieves (1) state-of-the-art zero-shot results on various video tasks using thousands of fewer GPU hours. (2) better performance than current video chatbots without any video instruction tuning. (3) state-of-the-art results of video chatting using video instruction tuning, outperforming previous SOTAs by a large margin.
Convolutional Networks with Oriented 1D Kernels
Authors: Alexandre Kirchmeyer, Jia Deng
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
In computer vision, 2D convolution is arguably the most important operation performed by a ConvNet. Unsurprisingly, it has been the focus of intense software and hardware optimization and enjoys highly efficient implementations. In this work, we ask an intriguing question: can we make a ConvNet work without 2D convolutions? Surprisingly, we find that the answer is yes -- we show that a ConvNet consisting entirely of 1D convolutions can do just as well as 2D on ImageNet classification. Specifically, we find that one key ingredient to a high-performing 1D ConvNet is oriented 1D kernels: 1D kernels that are oriented not just horizontally or vertically, but also at other angles. Our experiments show that oriented 1D convolutions can not only replace 2D convolutions but also augment existing architectures with large kernels, leading to improved accuracy with minimal FLOPs increase. A key contribution of this work is a highly-optimized custom CUDA implementation of oriented 1D kernels, specialized to the depthwise convolution setting. Our benchmarks demonstrate that our custom CUDA implementation almost perfectly realizes the theoretical advantage of 1D convolution: it is faster than a native horizontal convolution for any arbitrary angle. Code is available at https://github.com/princeton-vl/Oriented1D.
Keyword: mobile
Multimodal Dataset for Localization, Mapping and Crop Monitoring in Citrus Tree Farms
Abstract
In this work we introduce the CitrusFarm dataset, a comprehensive multimodal sensory dataset collected by a wheeled mobile robot operating in agricultural fields. The dataset offers stereo RGB images with depth information, as well as monochrome, near-infrared and thermal images, presenting diverse spectral responses crucial for agricultural research. Furthermore, it provides a range of navigational sensor data encompassing wheel odometry, LiDAR, inertial measurement unit (IMU), and GNSS with Real-Time Kinematic (RTK) as the centimeter-level positioning ground truth. The dataset comprises seven sequences collected in three fields of citrus trees, featuring various tree species at different growth stages, distinctive planting patterns, as well as varying daylight conditions. It spans a total operation time of 1.7 hours, covers a distance of 7.5 km, and constitutes 1.3 TB of data. We anticipate that this dataset can facilitate the development of autonomous robot systems operating in agricultural tree environments, especially for localization, mapping and crop monitoring tasks. Moreover, the rich sensing modalities offered in this dataset can also support research in a range of robotics and computer vision tasks, such as place recognition, scene understanding, object detection and segmentation, and multimodal learning. The dataset, in conjunction with related tools and resources, is made publicly available at https://github.com/UCR-Robotics/Citrus-Farm-Dataset.
Joint Computing, Pushing, and Caching Optimization for Mobile Edge Computing Networks via Soft Actor-Critic Learning
Abstract
Mobile edge computing (MEC) networks bring computing and storage capabilities closer to edge devices, which reduces latency and improves network performance. However, to further reduce transmission and computation costs while satisfying user-perceived quality of experience, a joint optimization in computing, pushing, and caching is needed. In this paper, we formulate the joint-design problem in MEC networks as an infinite-horizon discounted-cost Markov decision process and solve it using a deep reinforcement learning (DRL)-based framework that enables the dynamic orchestration of computing, pushing, and caching. Through the deep networks embedded in the DRL structure, our framework can implicitly predict user future requests and push or cache the appropriate content to effectively enhance system performance. One issue we encountered when considering three functions collectively is the curse of dimensionality for the action space. To address it, we relaxed the discrete action space into a continuous space and then adopted soft actor-critic learning to solve the optimization problem, followed by utilizing a vector quantization method to obtain the desired discrete action. Additionally, an action correction method was proposed to compress the action space further and accelerate the convergence. Our simulations under the setting of a general single-user, single-server MEC network with dynamic transmission link quality demonstrate that the proposed framework effectively decreases transmission bandwidth and computing cost by proactively pushing data on future demand to users and jointly optimizing the three functions. We also conduct extensive parameter tuning analysis, which shows that our approach outperforms the baselines under various parameter settings.
Teach and Repeat Navigation: A Robust Control Approach
Authors: Payam Nourizadeh, Michael Milford, Tobias Fischer
Subjects: Robotics (cs.RO); Systems and Control (eess.SY)
Abstract
Robot navigation requires an autonomy pipeline that is robust to environmental changes and effective in varying conditions. Teach and Repeat (T&R) navigation has shown high performance in autonomous repeated tasks under challenging circumstances, but research within T&R has predominantly focused on motion planning as opposed to motion control. In this paper, we propose a novel T&R system based on a robust motion control technique for a skid-steering mobile robot using sliding-mode control that effectively handles uncertainties that are particularly pronounced in the T&R task, where sensor noises, parametric uncertainties, and wheel-terrain interaction are common challenges. We first theoretically demonstrate that the proposed T&R system is globally stable and robust while considering the uncertainties of the closed-loop system. When deployed on a Clearpath Jackal robot, we then show the global stability of the proposed system in both indoor and outdoor environments covering different terrains, outperforming previous state-of-the-art methods in terms of mean average trajectory error and stability in these challenging environments. This paper makes an important step towards long-term autonomous T&R navigation with ensured safety guarantees.
GAMMA: Graspability-Aware Mobile MAnipulation Policy Learning based on Online Grasping Pose Fusion
Authors: Jiazhao Zhang, Nandiraju Gireesh, Jilong Wang, Xiaomeng Fang, Chaoyi Xu, Weiguang Chen, Liu Dai, He Wang
Subjects: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
Abstract
Mobile manipulation constitutes a fundamental task for robotic assistants and garners significant attention within the robotics community. A critical challenge inherent in mobile manipulation is the effective observation of the target while approaching it for grasping. In this work, we propose a graspability-aware mobile manipulation approach powered by an online grasping pose fusion framework that enables a temporally consistent grasping observation. Specifically, the predicted grasping poses are online organized to eliminate the redundant, outlier grasping poses, which can be encoded as a grasping pose observation state for reinforcement learning. Moreover, on-the-fly fusing the grasping poses enables a direct assessment of graspability, encompassing both the quantity and quality of grasping poses.
AdaEvo: Edge-Assisted Continuous and Timely DNN Model Evolution for Mobile Devices
Authors: Lehao Wang, Zhiwen Yu, Haoyi Yu, Sicong Liu, Yaxiong Xie, Bin Guo, Yunxin Liu
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC)
Abstract
Mobile video applications today have attracted significant attention. Deep learning model (e.g. deep neural network, DNN) compression is widely used to enable on-device inference for facilitating robust and private mobile video applications. The compressed DNN, however, is vulnerable to the agnostic data drift of the live video captured from the dynamically changing mobile scenarios. To combat the data drift, mobile ends rely on edge servers to continuously evolve and re-compress the DNN with freshly collected data. We design a framework, AdaEvo, that efficiently supports the resource-limited edge server handling mobile DNN evolution tasks from multiple mobile ends. The key goal of AdaEvo is to maximize the average quality of experience (QoE), e.g. the proportion of high-quality DNN service time to the entire life cycle, for all mobile ends. Specifically, it estimates the DNN accuracy drops at the mobile end without labels and performs a dedicated video frame sampling strategy to control the size of retraining data. In addition, it balances the limited computing and memory resources on the edge server and the competition between asynchronous tasks initiated by different mobile users. With an extensive evaluation of real-world videos from mobile scenarios and across four diverse mobile tasks, experimental results show that AdaEvo enables up to 34% accuracy improvement and 32% average QoE improvement.
Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models
Authors: Jung Hwan Heo, Jeonghoon Kim, Beomseok Kwon, Byeongwook Kim, Se Jung Kwon, Dongsoo Lee
Abstract
Large Language Models (LLMs) have recently demonstrated a remarkable success across various tasks. However, efficiently serving LLMs has been a challenge due to its large memory bottleneck, specifically in small batch inference settings (e.g. mobile devices). Weight-only quantization can be a promising approach, but sub-4 bit quantization remains a challenge due to large-magnitude activation outliers. To mitigate the undesirable outlier effect, we first propose per-IC quantization, a simple yet effective method that creates quantization groups within each input channel (IC) rather than the conventional per-output channel (OC). Our method is motivated by the observation that activation outliers affect the input dimension of the weight matrix, so similarly grouping the weights in the IC direction can isolate outliers to be within a group. We also find that activation outliers do not dictate quantization difficulty, and inherent weight sensitivities also exist. With per-IC quantization as a new outlier-friendly scheme, we then propose Adaptive Dimensions (AdaDim), a versatile quantization framework that can adapt to various weight sensitivity patterns. We demonstrate the effectiveness of AdaDim by augmenting prior methods such as Round-To-Nearest and GPTQ, showing significant improvements across various language modeling benchmarks for both base (up to +4.7% on MMLU) and instruction-tuned (up to +10% on HumanEval) LLMs.
Neuromorphic Imaging and Classification with Graph Learning
Authors: Pei Zhang, Chutian Wang, Edmund Y. Lam
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Bio-inspired neuromorphic cameras asynchronously record pixel brightness changes and generate sparse event streams. They can capture dynamic scenes with little motion blur and more details in extreme illumination conditions. Due to the multidimensional address-event structure, most existing vision algorithms cannot properly handle asynchronous event streams. While several event representations and processing methods have been developed to address such an issue, they are typically driven by a large number of events, leading to substantial overheads in runtime and memory. In this paper, we propose a new graph representation of the event data and couple it with a Graph Transformer to perform accurate neuromorphic classification. Extensive experiments show that our approach leads to better results and excels at the challenging realistic situations where only a small number of events and limited computational resources are available, paving the way for neuromorphic applications embedded into mobile facilities.
Tracking Snake-like Robots in the Wild Using Only a Single Camera
Authors: Jingpei Lu, Florian Richter, Shan Lin, Michael C. Yip
Abstract
Robot navigation within complex environments requires precise state estimation and localization to ensure robust and safe operations. For ambulating mobile robots like robot snakes, traditional methods for sensing require multiple embedded sensors or markers, leading to increased complexity, cost, and increased points of failure. Alternatively, deploying an external camera in the environment is very easy to do, and marker-less state estimation of the robot from this camera's images is an ideal solution: both simple and cost-effective. However, the challenge in this process is in tracking the robot under larger environments where the cameras may be moved around without extrinsic calibration, or maybe when in motion (e.g., a drone following the robot). The scenario itself presents a complex challenge: single-image reconstruction of robot poses under noisy observations. In this paper, we address the problem of tracking ambulatory mobile robots from a single camera. The method combines differentiable rendering with the Kalman filter. This synergy allows for simultaneous estimation of the robot's joint angle and pose while also providing state uncertainty which could be used later on for robust control. We demonstrate the efficacy of our approach on a snake-like robot in both stationary and non-stationary (moving) cameras, validating its performance in both structured and unstructured scenarios. The results achieved show an average error of 0.05 m in localizing the robot's base position and 6 degrees in joint state estimation. We believe this novel technique opens up possibilities for enhanced robot mobility and navigation in future exploratory and search-and-rescue missions.
Time-Domain Channel Measurements and Small-Scale Fading Characterization for RIS-Assisted Wireless Communication Systems
Abstract
As a potentially revolutionary enabling technology for the sixth generation (6G) mobile communication system, reconfigurable intelligent surfaces (RISs) have attracted extensive attention from industry and academia. In RIS-assisted wireless communication systems, practical channel measurements and modeling serve as the foundation for system design, network optimization, and performance evaluation. In this paper, a RIS time-domain channel measurement system, based on a software defined radio (SDR) platform, is developed for the first time to investigate the small-scale fading characteristics of RIS-assisted channels. We present RIS channel measurements in corridor and laboratory scenarios and compare the power delay profile (PDP) of the channel without RIS, with RIS specular reflection, and with RIS intelligent reflection. The multipath component parameters and cluster parameters based on the Saleh-Valenzuela model are extracted. We find that the PDPs of the RIS-assisted channel fit the power-law decay model and approximate the law of square decay. Through intelligent reflection, the RIS can decrease the delay and concentrate the energy of the virtual line-of-sight (VLOS) path, thereby reducing delay spread and mitigating multipath fading. Furthermore, the cluster characteristics of RIS-assisted channels are highly related to the measurement environment. In the laboratory scenario, a single cluster dominated by the VLOS path with smooth envelope is observed. On the other hand, in the corridor scenario, some additional clusters introduced by the RIS reflection are created.
Keyword: pruning
Model-Free, Regret-Optimal Best Policy Identification in Online CMDPs
Abstract
This paper considers the best policy identification (BPI) problem in online Constrained Markov Decision Processes (CMDPs). We are interested in algorithms that are model-free, have low regret, and identify an optimal policy with a high probability. Existing model-free algorithms for online CMDPs with sublinear regret and constraint violation do not provide any convergence guarantee to an optimal policy and provide only average performance guarantees when a policy is uniformly sampled at random from all previously used policies. In this paper, we develop a new algorithm, named Pruning-Refinement-Identification (PRI), based on a fundamental structural property of CMDPs we discover, called limited stochasticity. The property says for a CMDP with $N$ constraints, there exists an optimal policy with at most $N$ stochastic decisions. The proposed algorithm first identifies at which step and in which state a stochastic decision has to be taken and then fine-tunes the distributions of these stochastic decisions. PRI achieves trio objectives: (i) PRI is a model-free algorithm; and (ii) it outputs a near-optimal policy with a high probability at the end of learning; and (iii) in the tabular setting, PRI guarantees $\tilde{\mathcal{O}}(\sqrt{K})$ regret and constraint violation, which significantly improves the best existing regret bound $\tilde{\mathcal{O}}(K^{\frac{4}{5}})$ under a mode-free algorithm, where $K$ is the total number of episodes.
Abstract
The classic problem of exact subgraph matching returns those subgraphs in a large-scale data graph that are isomorphic to a given query graph, which has gained increasing importance in many real-world applications such as social network analysis, knowledge graph discovery in the Semantic Web, bibliographical network mining, and so on. In this paper, we propose a novel and effective \textit{graph neural network (GNN)-based path embedding framework} (GNN-PE), which allows efficient exact subgraph matching without introducing false dismissals. Unlike traditional GNN-based graph embeddings that only produce approximate subgraph matching results, in this paper, we carefully devise GNN-based embeddings for paths, such that: if two paths (and 1-hop neighbors of vertices on them) have the subgraph relationship, their corresponding GNN-based embedding vectors will strictly follow the dominance relationship. With such a newly designed property of path dominance embeddings, we are able to propose effective pruning strategies based on path label/dominance embeddings and guarantee no false dismissals for subgraph matching. We build multidimensional indexes over path embedding vectors, and develop an efficient subgraph matching algorithm by traversing indexes over graph partitions in parallel and applying our pruning methods. We also propose a cost-model-based query plan that obtains query paths from the query graph with low query cost. Through extensive experiments, we confirm the efficiency and effectiveness of our proposed GNN-PE approach for exact subgraph matching on both real and synthetic graph data.
CAIT: Triple-Win Compression towards High Accuracy, Fast Inference, and Favorable Transferability For ViTs
Abstract
Vision Transformers (ViTs) have emerged as state-of-the-art models for various vision tasks recently. However, their heavy computation costs remain daunting for resource-limited devices. Consequently, researchers have dedicated themselves to compressing redundant information in ViTs for acceleration. However, they generally sparsely drop redundant image tokens by token pruning or brutally remove channels by channel pruning, leading to a sub-optimal balance between model performance and inference speed. They are also disadvantageous in transferring compressed models to downstream vision tasks that require the spatial structure of images, such as semantic segmentation. To tackle these issues, we propose a joint compression method for ViTs that offers both high accuracy and fast inference speed, while also maintaining favorable transferability to downstream tasks (CAIT). Specifically, we introduce an asymmetric token merging (ATME) strategy to effectively integrate neighboring tokens. It can successfully compress redundant token information while preserving the spatial structure of images. We further employ a consistent dynamic channel pruning (CDCP) strategy to dynamically prune unimportant channels in ViTs. Thanks to CDCP, insignificant channels in multi-head self-attention modules of ViTs can be pruned uniformly, greatly enhancing the model compression. Extensive experiments on benchmark datasets demonstrate that our proposed method can achieve state-of-the-art performance across various ViTs. For example, our pruned DeiT-Tiny and DeiT-Small achieve speedups of 1.7$\times$ and 1.9$\times$, respectively, without accuracy drops on ImageNet. On the ADE20k segmentation dataset, our method can enjoy up to 1.31$\times$ speedups with comparable mIoU. Our code will be publicly available.
SHACIRA: Scalable HAsh-grid Compression for Implicit Neural Representations
Authors: Sharath Girish, Abhinav Shrivastava, Kamal Gupta
Abstract
Implicit Neural Representations (INR) or neural fields have emerged as a popular framework to encode multimedia signals such as images and radiance fields while retaining high-quality. Recently, learnable feature grids proposed by Instant-NGP have allowed significant speed-up in the training as well as the sampling of INRs by replacing a large neural network with a multi-resolution look-up table of feature vectors and a much smaller neural network. However, these feature grids come at the expense of large memory consumption which can be a bottleneck for storage and streaming applications. In this work, we propose SHACIRA, a simple yet effective task-agnostic framework for compressing such feature grids with no additional post-hoc pruning/quantization stages. We reparameterize feature grids with quantized latent weights and apply entropy regularization in the latent space to achieve high levels of compression across various domains. Quantitative and qualitative results on diverse datasets consisting of images, videos, and radiance fields, show that our approach outperforms existing INR approaches without the need for any large datasets or domain-specific heuristics. Our project page is available at this http URL .
Keyword: diffusion
PINF: Continuous Normalizing Flows for Physics-Constrained Deep Learning
Abstract
The normalization constraint on probability density poses a significant challenge for solving the Fokker-Planck equation. Normalizing Flow, an invertible generative model leverages the change of variables formula to ensure probability density conservation and enable the learning of complex data distributions. In this paper, we introduce Physics-Informed Normalizing Flows (PINF), a novel extension of continuous normalizing flows, incorporating diffusion through the method of characteristics. Our method, which is mesh-free and causality-free, can efficiently solve high dimensional time-dependent and steady-state Fokker-Planck equations.
Generative Residual Diffusion Modeling for Km-scale Atmospheric Downscaling
Abstract
The state of the art for physical hazard prediction from weather and climate requires expensive km-scale numerical simulations driven by coarser resolution global inputs. Here, a km-scale downscaling diffusion model is presented as a cost effective alternative. The model is trained from a regional high-resolution weather model over Taiwan, and conditioned on ERA5 reanalysis data. To address the downscaling uncertainties, large resolution ratios (25km to 2km), different physics involved at different scales and predict channels that are not in the input data, we employ a two-step approach (\textit{ResDiff}) where a (UNet) regression predicts the mean in the first step and a diffusion model predicts the residual in the second step. \textit{ResDiff} exhibits encouraging skill in bulk RMSE and CRPS scores. The predicted spectra and distributions from ResDiff faithfully recover important power law relationships regulating damaging wind and rain extremes. Case studies of coherent weather phenomena reveal appropriate multivariate relationships reminiscent of learnt physics. This includes the sharp wind and temperature variations that co-locate with intense rainfall in a cold front, and the extreme winds and rainfall bands that surround the eyewall of typhoons. Some evidence of simultaneous bias correction is found. A first attempt at downscaling directly from an operational global forecast model successfully retains many of these benefits. The implication is that a new era of fully end-to-end, global-to-regional machine learning weather prediction is likely near at hand.
Learning Using Generated Privileged Information by Text-to-Image Diffusion Models
Authors: Rafael-Edy Menadil, Mariana-Iuliana Georgescu, Radu Tudor Ionescu
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Abstract
Learning Using Privileged Information is a particular type of knowledge distillation where the teacher model benefits from an additional data representation during training, called privileged information, improving the student model, which does not see the extra representation. However, privileged information is rarely available in practice. To this end, we propose a text classification framework that harnesses text-to-image diffusion models to generate artificial privileged information. The generated images and the original text samples are further used to train multimodal teacher models based on state-of-the-art transformer-based architectures. Finally, the knowledge from multimodal teachers is distilled into a text-based (unimodal) student. Hence, by employing a generative model to produce synthetic data as privileged information, we guide the training of the student model. Our framework, called Learning Using Generated Privileged Information (LUGPI), yields noticeable performance gains on four text classification data sets, demonstrating its potential in text classification without any additional cost during inference.
Maximum Diffusion Reinforcement Learning
Authors: Thomas A. Berrueta, Allison Pinosky, Todd D. Murphey
Abstract
The assumption that data are independent and identically distributed underpins all machine learning. When data are collected sequentially from agent experiences this assumption does not generally hold, as in reinforcement learning. Here, we derive a method that overcomes these limitations by exploiting the statistical mechanics of ergodic processes, which we term maximum diffusion reinforcement learning. By decorrelating agent experiences, our approach provably enables agents to learn continually in single-shot deployments regardless of how they are initialized. Moreover, we prove our approach generalizes well-known maximum entropy techniques, and show that it robustly exceeds state-of-the-art performance across popular benchmarks. Our results at the nexus of physics, learning, and control pave the way towards more transparent and reliable decision-making in reinforcement learning agents, such as locomoting robots and self-driving cars.
LD4MRec: Simplifying and Powering Diffusion Model for Multimedia Recommendation
Authors: Penghang Yu, Zhiyi Tan, Guanming Lu, Bing-Kun Bao
Abstract
Multimedia recommendation aims to predict users' future behaviors based on historical behavioral data and item's multimodal information. However, noise inherent in behavioral data, arising from unintended user interactions with uninteresting items, detrimentally impacts recommendation performance. Recently, diffusion models have achieved high-quality information generation, in which the reverse process iteratively infers future information based on the corrupted state. It meets the need of predictive tasks under noisy conditions, and inspires exploring their application to predicting user behaviors. Nonetheless, several challenges must be addressed: 1) Classical diffusion models require excessive computation, which does not meet the efficiency requirements of recommendation systems. 2) Existing reverse processes are mainly designed for continuous data, whereas behavioral information is discrete in nature. Therefore, an effective method is needed for the generation of discrete behavioral information. To tackle the aforementioned issues, we propose a Light Diffusion model for Multimedia Recommendation. First, to reduce computational complexity, we simplify the formula of the reverse process, enabling one-step inference instead of multi-step inference. Second, to achieve effective behavioral information generation, we propose a novel Conditional neural Network. It maps the discrete behavior data into a continuous latent space, and generates behaviors with the guidance of collaborative signals and user multimodal preference. Additionally, considering that completely clean behavior data is inaccessible, we introduce a soft behavioral reconstruction constraint during model training, facilitating behavior prediction with noisy data. Empirical studies conducted on three public datasets demonstrate the effectiveness of LD4MRec.
Hypergraph $p$-Laplacians, Scale Spaces, and Information Flow in Networks
Authors: Ariane Fazeny, Daniel Tenbrinck, Martin Burger
Subjects: Social and Information Networks (cs.SI); Combinatorics (math.CO)
Abstract
This paper models opinion formation in social networks using oriented hypergraphs and it defines gradient flows in the form of diffusion equations on oriented hypergraphs. Therefore, this paper uses the gradient, adjoint and $p$-Laplacian definitions for oriented hypergraphs, introduced in arXiv:2304.06468v1, and applies them to modelling group dynamics and information flow in social networks.
DreamCom: Finetuning Text-guided Inpainting Model for Image Composition
Authors: Lingxiao Lu, Bo Zhang, Li Niu
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
The goal of image composition is merging a foreground object into a background image to obtain a realistic composite image. Recently, generative composition methods are built on large pretrained diffusion models, due to their unprecedented image generation ability. They train a model on abundant pairs of foregrounds and backgrounds, so that it can be directly applied to a new pair of foreground and background at test time. However, the generated results often lose the foreground details and exhibit noticeable artifacts. In this work, we propose an embarrassingly simple approach named DreamCom inspired by DreamBooth. Specifically, given a few reference images for a subject, we finetune text-guided inpainting diffusion model to associate this subject with a special token and inpaint this subject in the specified bounding box. We also construct a new dataset named MureCom well-tailored for this task.
High-Fidelity Speech Synthesis with Minimal Supervision: All Using Diffusion Models
Abstract
Text-to-speech (TTS) methods have shown promising results in voice cloning, but they require a large number of labeled text-speech pairs. Minimally-supervised speech synthesis decouples TTS by combining two types of discrete speech representations(semantic \& acoustic) and using two sequence-to-sequence tasks to enable training with minimal supervision. However, existing methods suffer from information redundancy and dimension explosion in semantic representation, and high-frequency waveform distortion in discrete acoustic representation. Autoregressive frameworks exhibit typical instability and uncontrollability issues. And non-autoregressive frameworks suffer from prosodic averaging caused by duration prediction models. To address these issues, we propose a minimally-supervised high-fidelity speech synthesis method, where all modules are constructed based on the diffusion models. The non-autoregressive framework enhances controllability, and the duration diffusion model enables diversified prosodic expression. Contrastive Token-Acoustic Pretraining (CTAP) is used as an intermediate semantic representation to solve the problems of information redundancy and dimension explosion in existing semantic coding methods. Mel-spectrogram is used as the acoustic representation. Both semantic and acoustic representations are predicted by continuous variable regression tasks to solve the problem of high-frequency fine-grained waveform distortion. Experimental results show that our proposed method outperforms the baseline method. We provide audio samples on our website.
Uncertainty Quantification via Neural Posterior Principal Components
Authors: Elias Nehme, Omer Yair, Tomer Michaeli
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV); Machine Learning (stat.ML)
Abstract
Uncertainty quantification is crucial for the deployment of image restoration models in safety-critical domains, like autonomous driving and biological imaging. To date, methods for uncertainty visualization have mainly focused on per-pixel estimates. However, a heatmap of per-pixel variances is typically of little practical use, as it does not capture the strong correlations between pixels. A more natural measure of uncertainty corresponds to the variances along the principal components (PCs) of the posterior distribution. Theoretically, the PCs can be computed by applying PCA on samples generated from a conditional generative model for the input image. However, this requires generating a very large number of samples at test time, which is painfully slow with the current state-of-the-art (diffusion) models. In this work, we present a method for predicting the PCs of the posterior distribution for any input image, in a single forward pass of a neural network. Our method can either wrap around a pre-trained model that was trained to minimize the mean square error (MSE), or can be trained from scratch to output both a predicted image and the posterior PCs. We showcase our method on multiple inverse problems in imaging, including denoising, inpainting, super-resolution, and biological image-to-image translation. Our method reliably conveys instance-adaptive uncertainty directions, achieving uncertainty quantification comparable with posterior samplers while being orders of magnitude faster. Examples are available at https://eliasnehme.github.io/NPPC/
Dynamic Prompt Learning: Addressing Cross-Attention Leakage for Text-Based Image Editing
Authors: Kai Wang, Fei Yang, Shiqi Yang, Muhammad Atif Butt, Joost van de Weijer
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Large-scale text-to-image generative models have been a ground-breaking development in generative AI, with diffusion models showing their astounding ability to synthesize convincing images following an input text prompt. The goal of image editing research is to give users control over the generated images by modifying the text prompt. Current image editing techniques are susceptible to unintended modifications of regions outside the targeted area, such as on the background or on distractor objects which have some semantic or visual relationship with the targeted object. According to our experimental findings, inaccurate cross-attention maps are at the root of this problem. Based on this observation, we propose Dynamic Prompt Learning (DPL) to force cross-attention maps to focus on correct noun words in the text prompt. By updating the dynamic tokens for nouns in the textual input with the proposed leakage repairment losses, we achieve fine-grained image editing over particular objects while preventing undesired changes to other image regions. Our method DPL, based on the publicly available Stable Diffusion, is extensively evaluated on a wide range of images, and consistently obtains superior results both quantitatively (CLIP score, Structure-Dist) and qualitatively (on user-evaluation). We show improved prompt editing results for Word-Swap, Prompt Refinement, and Attention Re-weighting, especially for complex multi-object scenes.
Factorized Diffusion Architectures for Unsupervised Image Generation and Segmentation
Authors: Xin Yuan, Michael Maire
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
We develop a neural network architecture which, trained in an unsupervised manner as a denoising diffusion model, simultaneously learns to both generate and segment images. Learning is driven entirely by the denoising diffusion objective, without any annotation or prior knowledge about regions during training. A computational bottleneck, built into the neural architecture, encourages the denoising network to partition an input into regions, denoise them in parallel, and combine the results. Our trained model generates both synthetic images and, by simple examination of its internal predicted partitions, a semantic segmentation of those images. Without any finetuning, we directly apply our unsupervised model to the downstream task of segmenting real images via noising and subsequently denoising them. Experiments demonstrate that our model achieves accurate unsupervised image segmentation and high-quality synthetic image generation across multiple datasets.
Emu: Enhancing Image Generation Models Using Photogenic Needles in a Haystack
Authors: Xiaoliang Dai, Ji Hou, Chih-Yao Ma, Sam Tsai, Jialiang Wang, Rui Wang, Peizhao Zhang, Simon Vandenhende, Xiaofang Wang, Abhimanyu Dubey, Matthew Yu, Abhishek Kadian, Filip Radenovic, Dhruv Mahajan, Kunpeng Li, Yue Zhao, Vladan Petrovic, Mitesh Kumar Singh, Simran Motwani, Yi Wen, Yiwen Song, Roshan Sumbaly, Vignesh Ramanathan, Zijian He, Peter Vajda, Devi Parikh
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Training text-to-image models with web scale image-text pairs enables the generation of a wide range of visual concepts from text. However, these pre-trained models often face challenges when it comes to generating highly aesthetic images. This creates the need for aesthetic alignment post pre-training. In this paper, we propose quality-tuning to effectively guide a pre-trained model to exclusively generate highly visually appealing images, while maintaining generality across visual concepts. Our key insight is that supervised fine-tuning with a set of surprisingly small but extremely visually appealing images can significantly improve the generation quality. We pre-train a latent diffusion model on $1.1$ billion image-text pairs and fine-tune it with only a few thousand carefully selected high-quality images. The resulting model, Emu, achieves a win rate of $82.9\%$ compared with its pre-trained only counterpart. Compared to the state-of-the-art SDXLv1.0, Emu is preferred $68.4\%$ and $71.3\%$ of the time on visual appeal on the standard PartiPrompts and our Open User Input benchmark based on the real-world usage of text-to-image models. In addition, we show that quality-tuning is a generic approach that is also effective for other architectures, including pixel diffusion and masked generative transformer models.
Show-1: Marrying Pixel and Latent Diffusion Models for Text-to-Video Generation
Authors: David Junhao Zhang, Jay Zhangjie Wu, Jia-Wei Liu, Rui Zhao, Lingmin Ran, Yuchao Gu, Difei Gao, Mike Zheng Shou
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Significant advancements have been achieved in the realm of large-scale pre-trained text-to-video Diffusion Models (VDMs). However, previous methods either rely solely on pixel-based VDMs, which come with high computational costs, or on latent-based VDMs, which often struggle with precise text-video alignment. In this paper, we are the first to propose a hybrid model, dubbed as Show-1, which marries pixel-based and latent-based VDMs for text-to-video generation. Our model first uses pixel-based VDMs to produce a low-resolution video of strong text-video correlation. After that, we propose a novel expert translation method that employs the latent-based VDMs to further upsample the low-resolution video to high resolution. Compared to latent VDMs, Show-1 can produce high-quality videos of precise text-video alignment; Compared to pixel VDMs, Show-1 is much more efficient (GPU memory usage during inference is 15G vs 72G). We also validate our model on standard video generation benchmarks. Our code and model weights are publicly available at \url{https://github.com/showlab/Show-1}.
LGMCTS: Language-Guided Monte-Carlo Tree Search for Executable Semantic Object Rearrangement
Authors: Haonan Chang, Kai Gao, Kowndinya Boyalakuntla, Alex Lee, Baichuan Huang, Harish Udhaya Kumar, Jinjin Yu, Abdeslam Boularias
Abstract
We introduce a novel approach to the executable semantic object rearrangement problem. In this challenge, a robot seeks to create an actionable plan that rearranges objects within a scene according to a pattern dictated by a natural language description. Unlike existing methods such as StructFormer and StructDiffusion, which tackle the issue in two steps by first generating poses and then leveraging a task planner for action plan formulation, our method concurrently addresses pose generation and action planning. We achieve this integration using a Language-Guided Monte-Carlo Tree Search (LGMCTS). Quantitative evaluations are provided on two simulation datasets, and complemented by qualitative tests with a real robot.
Exploiting the Signal-Leak Bias in Diffusion Models
Authors: Martin Nicolas Everaert, Athanasios Fitsios, Marco Bocchio, Sami Arpa, Sabine Süsstrunk, Radhakrishna Achanta
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
There is a bias in the inference pipeline of most diffusion models. This bias arises from a signal leak whose distribution deviates from the noise distribution, creating a discrepancy between training and inference processes. We demonstrate that this signal-leak bias is particularly significant when models are tuned to a specific style, causing sub-optimal style matching. Recent research tries to avoid the signal leakage during training. We instead show how we can exploit this signal-leak bias in existing diffusion models to allow more control over the generated images. This enables us to generate images with more varied brightness, and images that better match a desired style or color. By modeling the distribution of the signal leak in the spatial frequency and pixel domains, and including a signal leak in the initial latent, we generate images that better match expected results without any additional training.
Keyword: adaptive
STANCE-C3: Domain-adaptive Cross-target Stance Detection via Contrastive Learning and Counterfactual Generation
Authors: Nayoung Kim, David Mosallanezhad, Lu Cheng, Michelle V. Mancenido, Huan Liu
Subjects: Computation and Language (cs.CL); Social and Information Networks (cs.SI)
Abstract
Stance detection is the process of inferring a person's position or standpoint on a specific issue to deduce prevailing perceptions toward topics of general or controversial interest, such as health policies during the COVID-19 pandemic. Existing models for stance detection are trained to perform well for a single domain (e.g., COVID-19) and a specific target topic (e.g., masking protocols), but are generally ineffectual in other domains or targets due to distributional shifts in the data. However, constructing high-performing, domain-specific stance detection models requires an extensive corpus of labeled data relevant to the targeted domain, yet such datasets are not readily available. This poses a challenge as the process of annotating data is costly and time-consuming. To address these challenges, we introduce a novel stance detection model coined domain-adaptive Cross-target STANCE detection via Contrastive learning and Counterfactual generation (STANCE-C3) that uses counterfactual data augmentation to enhance domain-adaptive training by enriching the target domain dataset during the training process and requiring significantly less information from the new domain. We also propose a modified self-supervised contrastive learning as a component of STANCE-C3 to prevent overfitting for the existing domain and target and enable cross-target stance detection. Through experiments on various datasets, we show that STANCE-C3 shows performance improvement over existing state-of-the-art methods.
Balancing Computational Efficiency and Forecast Error in Machine Learning-based Time-Series Forecasting: Insights from Live Experiments on Meteorological Nowcasting
Authors: Elin Törnquist, Wagner Costa Santos, Timothy Pogue, Nicholas Wingle, Robert A. Caulk
Abstract
Machine learning for time-series forecasting remains a key area of research. Despite successful application of many machine learning techniques, relating computational efficiency to forecast error remains an under-explored domain. This paper addresses this topic through a series of real-time experiments to quantify the relationship between computational cost and forecast error using meteorological nowcasting as an example use-case. We employ a variety of popular regression techniques (XGBoost, FC-MLP, Transformer, and LSTM) for multi-horizon, short-term forecasting of three variables (temperature, wind speed, and cloud cover) for multiple locations. During a 5-day live experiment, 4000 data sources were streamed for training and inferencing 144 models per hour. These models were parameterized to explore forecast error for two computational cost minimization methods: a novel auto-adaptive data reduction technique (Variance Horizon) and a performance-based concept drift-detection mechanism. Forecast error of all model variations were benchmarked in real-time against a state-of-the-art numerical weather prediction model. Performance was assessed using classical and novel evaluation metrics. Results indicate that using the Variance Horizon reduced computational usage by more than 50\%, while increasing between 0-15\% in error. Meanwhile, performance-based retraining reduced computational usage by up to 90\% while \emph{also} improving forecast error by up to 10\%. Finally, the combination of both the Variance Horizon and performance-based retraining outperformed other model configurations by up to 99.7\% when considering error normalized to computational usage.
Scaling Representation Learning from Ubiquitous ECG with State-Space Models
Abstract
Ubiquitous sensing from wearable devices in the wild holds promise for enhancing human well-being, from diagnosing clinical conditions and measuring stress to building adaptive health promoting scaffolds. But the large volumes of data therein across heterogeneous contexts pose challenges for conventional supervised learning approaches. Representation Learning from biological signals is an emerging realm catalyzed by the recent advances in computational modeling and the abundance of publicly shared databases. The electrocardiogram (ECG) is the primary researched modality in this context, with applications in health monitoring, stress and affect estimation. Yet, most studies are limited by small-scale controlled data collection and over-parameterized architecture choices. We introduce \textbf{WildECG}, a pre-trained state-space model for representation learning from ECG signals. We train this model in a self-supervised manner with 275,000 10s ECG recordings collected in the wild and evaluate it on a range of downstream tasks. The proposed model is a robust backbone for ECG analysis, providing competitive performance on most of the tasks considered, while demonstrating efficacy in low-resource regimes. The code and pre-trained weights are shared publicly at https://github.com/klean2050/tiles_ecg_model.
Seeing Beyond the Patch: Scale-Adaptive Semantic Segmentation of High-resolution Remote Sensing Imagery based on Reinforcement Learning
Abstract
In remote sensing imagery analysis, patch-based methods have limitations in capturing information beyond the sliding window. This shortcoming poses a significant challenge in processing complex and variable geo-objects, which results in semantic inconsistency in segmentation results. To address this challenge, we propose a dynamic scale perception framework, named GeoAgent, which adaptively captures appropriate scale context information outside the image patch based on the different geo-objects. In GeoAgent, each image patch's states are represented by a global thumbnail and a location mask. The global thumbnail provides context beyond the patch, and the location mask guides the perceived spatial relationships. The scale-selection actions are performed through a Scale Control Agent (SCA). A feature indexing module is proposed to enhance the ability of the agent to distinguish the current image patch's location. The action switches the patch scale and context branch of a dual-branch segmentation network that extracts and fuses the features of multi-scale patches. The GeoAgent adjusts the network parameters to perform the appropriate scale-selection action based on the reward received for the selected scale. The experimental results, using two publicly available datasets and our newly constructed dataset WUSU, demonstrate that GeoAgent outperforms previous segmentation methods, particularly for large-scale mapping applications.
MINS: Efficient and Robust Multisensor-aided Inertial Navigation System
Authors: Woosik Lee, Patrick Geneva, Chuchu Chen, Guoquan Huang
Abstract
Robust multisensor fusion of multi-modal measurements such as IMUs, wheel encoders, cameras, LiDARs, and GPS holds great potential due to its innate ability to improve resilience to sensor failures and measurement outliers, thereby enabling robust autonomy. To the best of our knowledge, this work is among the first to develop a consistent tightly-coupled Multisensor-aided Inertial Navigation System (MINS) that is capable of fusing the most common navigation sensors in an efficient filtering framework, by addressing the particular challenges of computational complexity, sensor asynchronicity, and intra-sensor calibration. In particular, we propose a consistent high-order on-manifold interpolation scheme to enable efficient asynchronous sensor fusion and state management strategy (i.e. dynamic cloning). The proposed dynamic cloning leverages motion-induced information to adaptively select interpolation orders to control computational complexity while minimizing trajectory representation errors. We perform online intrinsic and extrinsic (spatiotemporal) calibration of all onboard sensors to compensate for poor prior calibration and/or degraded calibration varying over time. Additionally, we develop an initialization method with only proprioceptive measurements of IMU and wheel encoders, instead of exteroceptive sensors, which is shown to be less affected by the environment and more robust in highly dynamic scenarios. We extensively validate the proposed MINS in simulations and large-scale challenging real-world datasets, outperforming the existing state-of-the-art methods, in terms of localization accuracy, consistency, and computation efficiency. We have also open-sourced our algorithm, simulator, and evaluation toolbox for the benefit of the community: https://github.com/rpng/mins.
Parallel local time stepping for rigid bodies represented by triangulated meshes
Abstract
Discrete Element Methods (DEM), i.e.~the simulation of many rigid particles, suffer from very stiff differential equations plus multiscale challenges in space and time. The particles move smoothly through space until they interact almost instantaneously due to collisions. Dense particle packings hence require tiny time step sizes, while free particles can advance with large time steps. Admissible time step sizes can span multiple orders of magnitudes. We propose an adaptive local time stepping algorithm which identifies clusters of particles that can be updated independently, advances them optimistically and independently in time, determines collision time stamps in space-time such that we maximise the time step sizes used, and resolves the momentum exchange implicitly. It is combined with various acceleration techniques which exploit multiscale geometry representations and multiscale behaviour in time. The collision time stamp detection in space-time in combination with the implicit solve of the actual collision equations avoids that particles get locked into tiny time step sizes, the clustering yields a high concurrency level, and the acceleration techniques plus local time stepping avoid unnecessary computations. This brings a scaling, adaptive time stepping for DEM for real-world challenges into reach.
Automatic Feature Fairness in Recommendation via Adversaries
Abstract
Fairness is a widely discussed topic in recommender systems, but its practical implementation faces challenges in defining sensitive features while maintaining recommendation accuracy. We propose feature fairness as the foundation to achieve equitable treatment across diverse groups defined by various feature combinations. This improves overall accuracy through balanced feature generalizability. We introduce unbiased feature learning through adversarial training, using adversarial perturbation to enhance feature representation. The adversaries improve model generalization for under-represented features. We adapt adversaries automatically based on two forms of feature biases: frequency and combination variety of feature values. This allows us to dynamically adjust perturbation strengths and adversarial training weights. Stronger perturbations are applied to feature values with fewer combination varieties to improve generalization, while higher weights for low-frequency features address training imbalances. We leverage the Adaptive Adversarial perturbation based on the widely-applied Factorization Machine (AAFM) as our backbone model. In experiments, AAFM surpasses strong baselines in both fairness and accuracy measures. AAFM excels in providing item- and user-fairness for single- and multi-feature tasks, showcasing their versatility and scalability. To maintain good accuracy, we find that adversarial perturbation must be well-managed: during training, perturbations should not overly persist and their strengths should decay.
NeuRBF: A Neural Fields Representation with Adaptive Radial Basis Functions
Abstract
We present a novel type of neural fields that uses general radial bases for signal representation. State-of-the-art neural fields typically rely on grid-based representations for storing local neural features and N-dimensional linear kernels for interpolating features at continuous query points. The spatial positions of their neural features are fixed on grid nodes and cannot well adapt to target signals. Our method instead builds upon general radial bases with flexible kernel position and shape, which have higher spatial adaptivity and can more closely fit target signals. To further improve the channel-wise capacity of radial basis functions, we propose to compose them with multi-frequency sinusoid functions. This technique extends a radial basis to multiple Fourier radial bases of different frequency bands without requiring extra parameters, facilitating the representation of details. Moreover, by marrying adaptive radial bases with grid-based ones, our hybrid combination inherits both adaptivity and interpolation smoothness. We carefully designed weighting schemes to let radial bases adapt to different types of signals effectively. Our experiments on 2D image and 3D signed distance field representation demonstrate the higher accuracy and compactness of our method than prior arts. When applied to neural radiance field reconstruction, our method achieves state-of-the-art rendering quality, with small model size and comparable training speed.
Cardinality Estimation of Subgraph Matching: A Filtering-Sampling Approach
Authors: Wonseok Shin, Siwoo Song, Kunsoo Park, Wook-Shin Han
Abstract
Subgraph counting is a fundamental problem in understanding and analyzing graph structured data, yet computationally challenging. This calls for an accurate and efficient algorithm for Subgraph Cardinality Estimation, which is to estimate the number of all isomorphic embeddings of a query graph in a data graph. We present FaSTest, a novel algorithm that combines (1) a powerful filtering technique to significantly reduce the sample space, (2) an adaptive tree sampling algorithm for accurate and efficient estimation, and (3) a worst-case optimal stratified graph sampling algorithm for difficult instances. Extensive experiments on real-world datasets show that FaSTest outperforms state-of-the-art sampling-based methods by up to two orders of magnitude and GNN-based methods by up to three orders of magnitude in terms of accuracy.
Enabling Resource-efficient AIoT System with Cross-level Optimization: A survey
Abstract
The emerging field of artificial intelligence of things (AIoT, AI+IoT) is driven by the widespread use of intelligent infrastructures and the impressive success of deep learning (DL). With the deployment of DL on various intelligent infrastructures featuring rich sensors and weak DL computing capabilities, a diverse range of AIoT applications has become possible. However, DL models are notoriously resource-intensive. Existing research strives to realize near-/realtime inference of AIoT live data and low-cost training using AIoT datasets on resource-scare infrastructures. Accordingly, the accuracy and responsiveness of DL models are bounded by resource availability. To this end, the algorithm-system co-design that jointly optimizes the resource-friendly DL models and model-adaptive system scheduling improves the runtime resource availability and thus pushes the performance boundary set by the standalone level. Unlike previous surveys on resource-friendly DL models or hand-crafted DL compilers/frameworks with partially fine-tuned components, this survey aims to provide a broader optimization space for more free resource-performance tradeoffs. The cross-level optimization landscape involves various granularity, including the DL model, computation graph, operator, memory schedule, and hardware instructor in both on-device and distributed paradigms. Furthermore, due to the dynamic nature of AIoT context, which includes heterogeneous hardware, agnostic sensing data, varying user-specified performance demands, and resource constraints, this survey explores the context-aware inter-/intra-device controllers for automatic cross-level adaptation. Additionally, we identify some potential directions for resource-efficient AIoT systems. By consolidating problems and techniques scattered over diverse levels, we aim to help readers understand their connections and stimulate further discussions.
Towards Human-Like RL: Taming Non-Naturalistic Behavior in Deep RL via Adaptive Behavioral Costs in 3D Games
Abstract
In this paper, we propose a new approach called Adaptive Behavioral Costs in Reinforcement Learning (ABC-RL) for training a human-like agent with competitive strength. While deep reinforcement learning agents have recently achieved superhuman performance in various video games, some of these unconstrained agents may exhibit actions, such as shaking and spinning, that are not typically observed in human behavior, resulting in peculiar gameplay experiences. To behave like humans and retain similar performance, ABC-RL augments behavioral limitations as cost signals in reinforcement learning with dynamically adjusted weights. Unlike traditional constrained policy optimization, we propose a new formulation that minimizes the behavioral costs subject to a constraint of the value function. By leveraging the augmented Lagrangian, our approach is an approximation of the Lagrangian adjustment, which handles the trade-off between the performance and the human-like behavior. Through experiments conducted on 3D games in DMLab-30 and Unity ML-Agents Toolkit, we demonstrate that ABC-RL achieves the same performance level while significantly reducing instances of shaking and spinning. These findings underscore the effectiveness of our proposed approach in promoting more natural and human-like behavior during gameplay.
Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models
Authors: Jung Hwan Heo, Jeonghoon Kim, Beomseok Kwon, Byeongwook Kim, Se Jung Kwon, Dongsoo Lee
Abstract
Large Language Models (LLMs) have recently demonstrated a remarkable success across various tasks. However, efficiently serving LLMs has been a challenge due to its large memory bottleneck, specifically in small batch inference settings (e.g. mobile devices). Weight-only quantization can be a promising approach, but sub-4 bit quantization remains a challenge due to large-magnitude activation outliers. To mitigate the undesirable outlier effect, we first propose per-IC quantization, a simple yet effective method that creates quantization groups within each input channel (IC) rather than the conventional per-output channel (OC). Our method is motivated by the observation that activation outliers affect the input dimension of the weight matrix, so similarly grouping the weights in the IC direction can isolate outliers to be within a group. We also find that activation outliers do not dictate quantization difficulty, and inherent weight sensitivities also exist. With per-IC quantization as a new outlier-friendly scheme, we then propose Adaptive Dimensions (AdaDim), a versatile quantization framework that can adapt to various weight sensitivity patterns. We demonstrate the effectiveness of AdaDim by augmenting prior methods such as Round-To-Nearest and GPTQ, showing significant improvements across various language modeling benchmarks for both base (up to +4.7% on MMLU) and instruction-tuned (up to +10% on HumanEval) LLMs.
Uncertainty Quantification via Neural Posterior Principal Components
Authors: Elias Nehme, Omer Yair, Tomer Michaeli
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV); Machine Learning (stat.ML)
Abstract
Uncertainty quantification is crucial for the deployment of image restoration models in safety-critical domains, like autonomous driving and biological imaging. To date, methods for uncertainty visualization have mainly focused on per-pixel estimates. However, a heatmap of per-pixel variances is typically of little practical use, as it does not capture the strong correlations between pixels. A more natural measure of uncertainty corresponds to the variances along the principal components (PCs) of the posterior distribution. Theoretically, the PCs can be computed by applying PCA on samples generated from a conditional generative model for the input image. However, this requires generating a very large number of samples at test time, which is painfully slow with the current state-of-the-art (diffusion) models. In this work, we present a method for predicting the PCs of the posterior distribution for any input image, in a single forward pass of a neural network. Our method can either wrap around a pre-trained model that was trained to minimize the mean square error (MSE), or can be trained from scratch to output both a predicted image and the posterior PCs. We showcase our method on multiple inverse problems in imaging, including denoising, inpainting, super-resolution, and biological image-to-image translation. Our method reliably conveys instance-adaptive uncertainty directions, achieving uncertainty quantification comparable with posterior samplers while being orders of magnitude faster. Examples are available at https://eliasnehme.github.io/NPPC/
Keyword: quantization
Joint Computing, Pushing, and Caching Optimization for Mobile Edge Computing Networks via Soft Actor-Critic Learning
Abstract
Mobile edge computing (MEC) networks bring computing and storage capabilities closer to edge devices, which reduces latency and improves network performance. However, to further reduce transmission and computation costs while satisfying user-perceived quality of experience, a joint optimization in computing, pushing, and caching is needed. In this paper, we formulate the joint-design problem in MEC networks as an infinite-horizon discounted-cost Markov decision process and solve it using a deep reinforcement learning (DRL)-based framework that enables the dynamic orchestration of computing, pushing, and caching. Through the deep networks embedded in the DRL structure, our framework can implicitly predict user future requests and push or cache the appropriate content to effectively enhance system performance. One issue we encountered when considering three functions collectively is the curse of dimensionality for the action space. To address it, we relaxed the discrete action space into a continuous space and then adopted soft actor-critic learning to solve the optimization problem, followed by utilizing a vector quantization method to obtain the desired discrete action. Additionally, an action correction method was proposed to compress the action space further and accelerate the convergence. Our simulations under the setting of a general single-user, single-server MEC network with dynamic transmission link quality demonstrate that the proposed framework effectively decreases transmission bandwidth and computing cost by proactively pushing data on future demand to users and jointly optimizing the three functions. We also conduct extensive parameter tuning analysis, which shows that our approach outperforms the baselines under various parameter settings.
Finite Scalar Quantization: VQ-VAE Made Simple
Authors: Fabian Mentzer, David Minnen, Eirikur Agustsson, Michael Tschannen
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
We propose to replace vector quantization (VQ) in the latent representation of VQ-VAEs with a simple scheme termed finite scalar quantization (FSQ), where we project the VAE representation down to a few dimensions (typically less than 10). Each dimension is quantized to a small set of fixed values, leading to an (implicit) codebook given by the product of these sets. By appropriately choosing the number of dimensions and values each dimension can take, we obtain the same codebook size as in VQ. On top of such discrete representations, we can train the same models that have been trained on VQ-VAE representations. For example, autoregressive and masked transformer models for image generation, multimodal generation, and dense prediction computer vision tasks. Concretely, we employ FSQ with MaskGIT for image generation, and with UViM for depth estimation, colorization, and panoptic segmentation. Despite the much simpler design of FSQ, we obtain competitive performance in all these tasks. We emphasize that FSQ does not suffer from codebook collapse and does not need the complex machinery employed in VQ (commitment losses, codebook reseeding, code splitting, entropy penalties, etc.) to learn expressive discrete representations.
Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models
Authors: Jung Hwan Heo, Jeonghoon Kim, Beomseok Kwon, Byeongwook Kim, Se Jung Kwon, Dongsoo Lee
Abstract
Large Language Models (LLMs) have recently demonstrated a remarkable success across various tasks. However, efficiently serving LLMs has been a challenge due to its large memory bottleneck, specifically in small batch inference settings (e.g. mobile devices). Weight-only quantization can be a promising approach, but sub-4 bit quantization remains a challenge due to large-magnitude activation outliers. To mitigate the undesirable outlier effect, we first propose per-IC quantization, a simple yet effective method that creates quantization groups within each input channel (IC) rather than the conventional per-output channel (OC). Our method is motivated by the observation that activation outliers affect the input dimension of the weight matrix, so similarly grouping the weights in the IC direction can isolate outliers to be within a group. We also find that activation outliers do not dictate quantization difficulty, and inherent weight sensitivities also exist. With per-IC quantization as a new outlier-friendly scheme, we then propose Adaptive Dimensions (AdaDim), a versatile quantization framework that can adapt to various weight sensitivity patterns. We demonstrate the effectiveness of AdaDim by augmenting prior methods such as Round-To-Nearest and GPTQ, showing significant improvements across various language modeling benchmarks for both base (up to +4.7% on MMLU) and instruction-tuned (up to +10% on HumanEval) LLMs.
Low Latency of object detection for spikng neural network
Authors: Nemin Qiu, Chuang Zhu
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Spiking Neural Networks, as a third-generation neural network, are well-suited for edge AI applications due to their binary spike nature. However, when it comes to complex tasks like object detection, SNNs often require a substantial number of time steps to achieve high performance. This limitation significantly hampers the widespread adoption of SNNs in latency-sensitive edge devices. In this paper, our focus is on generating highly accurate and low-latency SNNs specifically for object detection. Firstly, we systematically derive the conversion between SNNs and ANNs and analyze how to improve the consistency between them: improving the spike firing rate and reducing the quantization error. Then we propose a structural replacement, quantization of ANN activation and residual fix to allevicate the disparity. We evaluate our method on challenging dataset MS COCO, PASCAL VOC and our spike dataset. The experimental results show that the proposed method achieves higher accuracy and lower latency compared to previous work Spiking-YOLO. The advantages of SNNs processing of spike signals are also demonstrated.
SHACIRA: Scalable HAsh-grid Compression for Implicit Neural Representations
Authors: Sharath Girish, Abhinav Shrivastava, Kamal Gupta
Abstract
Implicit Neural Representations (INR) or neural fields have emerged as a popular framework to encode multimedia signals such as images and radiance fields while retaining high-quality. Recently, learnable feature grids proposed by Instant-NGP have allowed significant speed-up in the training as well as the sampling of INRs by replacing a large neural network with a multi-resolution look-up table of feature vectors and a much smaller neural network. However, these feature grids come at the expense of large memory consumption which can be a bottleneck for storage and streaming applications. In this work, we propose SHACIRA, a simple yet effective task-agnostic framework for compressing such feature grids with no additional post-hoc pruning/quantization stages. We reparameterize feature grids with quantized latent weights and apply entropy regularization in the latent space to achieve high levels of compression across various domains. Quantitative and qualitative results on diverse datasets consisting of images, videos, and radiance fields, show that our approach outperforms existing INR approaches without the need for any large datasets or domain-specific heuristics. Our project page is available at this http URL .
Keyword: efficient
PINF: Continuous Normalizing Flows for Physics-Constrained Deep Learning
A Review on AI Algorithms for Energy Management in E-Mobility Services
3D Reconstruction with Generalizable Neural Fields using Scene Priors
Auto-grading C programming assignments with CodeBERT and Random Forest Regressor
VPA: Fully Test-Time Visual Prompt Adaptation
Memory-Efficient Continual Learning Object Segmentation for Long Video
Efficient Low-rank Backpropagation for Vision Transformer Adaptation
joint prediction and denoising for large-scale multilingual self-supervised learning
Beyond the Chat: Executable and Verifiable Text-Editing with LLMs
Multi-dimensional Data Quick Query for Blockchain-based Federated Learning
Human-robot Matching and Routing for Multi-robot Tour Guiding under Time Uncertainty
PPG to ECG Signal Translation for Continuous Atrial Fibrillation Detection via Attention-based Deep State-Space Modeling
Adversarial Object Rearrangement in Constrained Environments with Heterogeneous Graph Neural Networks
Intelligent trading strategy based on improved directional change and regime change detection
MINS: Efficient and Robust Multisensor-aided Inertial Navigation System
KDD-LOAM: Jointly Learned Keypoint Detector and Descriptors Assisted LiDAR Odometry and Mapping
SOCI^+: An Enhanced Toolkit for Secure OutsourcedComputation on Integers
Prosumers Participation in Markets: A Scalar-Parameterized Function Bidding Approach
Cardinality Estimation of Subgraph Matching: A Filtering-Sampling Approach
Semantics-Driven Cloud-Edge Collaborative Inference
In-Hand Re-grasp Manipulation with Passive Dynamic Actions via Imitation Learning
LogicMP: A Neuro-symbolic Approach for Encoding First-order Logic Constraints
Enabling Resource-efficient AIoT System with Cross-level Optimization: A survey
Patterns Induce Injectivity: A New Thinking in Constructing Injective Local Rules of 1D Cellular Automata over $\mathbb{F}_2$
Fast Locality Sensitive Hashing with Theoretical Guarantee
Energy-Efficient Precoding Designs for Multi-User Visible Light Communication Systems with Confidential Messages
VideoAdviser: Video Knowledge Distillation for Multimodal Transfer Learning
AdaEvo: Edge-Assisted Continuous and Timely DNN Model Evolution for Mobile Devices
Overcoming the Fear of the Dark: Occlusion-Aware Model-Predictive Planning for Automated Vehicles Using Risk Fields
Teaching Text-to-Image Models to Communicate
Residual Scheduling: A New Reinforcement Learning Approach to Solving Job Shop Scheduling Problem
Raijū: Reinforcement Learning-Guided Post-Exploitation for Automating Security Assessment of Network Systems
Defending Against Physical Adversarial Patch Attacks on Infrared Human Detection
SAF-Net: Self-Attention Fusion Network for Myocardial Infarction Detection using Multi-View Echocardiography
Improving Facade Parsing with Vision Transformers and Line Integration
Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models
Towards Faithful Neural Network Intrinsic Interpretation with Shapley Additive Self-Attribution
Jointly Training Large Autoregressive Multimodal Models
HPL-ViT: A Unified Perception Framework for Heterogeneous Parallel LiDARs in V2V
Advancing the distributed Multi-GPU ChASE library through algorithm optimization and NCCL library
PolarNet: 3D Point Clouds for Language-Guided Robotic Manipulation
RapidEarth: A Search-by-Classification Engine for Large-Scale Geospatial Imagery
Design and Optimization of Residual Neural Network Accelerators for Low-Power FPGAs Using High-Level Synthesis
Efficient Exact Subgraph Matching via GNN-based Path Dominance Embedding (Technical Report)
Tactile-based Active Inference for Force-Controlled Peg-in-Hole Insertions
Improving Autonomous Driving Safety with POP: A Framework for Accurate Partially Observed Trajectory Predictions
Physics-Based Rigid Body Object Tracking and Friction Filtering From RGB-D Videos
MindGPT: Interpreting What You See with Non-invasive Brain Recordings
Generating Transferable Adversarial Simulation Scenarios for Self-Driving via Neural Rendering
Learning the Efficient Frontier
Partial Transport for Point-Cloud Registration
Convolutional Networks with Oriented 1D Kernels
Show-1: Marrying Pixel and Latent Diffusion Models for Text-to-Video Generation
OrthoPlanes: A Novel Representation for Better 3D-Awareness of GANs
Keyword: faster
RAGAS: Automated Evaluation of Retrieval Augmented Generation
A Physics Enhanced Residual Learning (PERL) Framework for Traffic State Prediction
MAPTree: Beating "Optimal" Decision Trees with Bayesian Decision Trees
Neural Operators for Accelerating Scientific Simulations and Design
Semantics-Driven Cloud-Edge Collaborative Inference
P2I-NET: Mapping Camera Pose to Image via Adversarial Learning for New View Synthesis in Real Indoor Environments
Uncertainty Quantification via Neural Posterior Principal Components
Rapid Network Adaptation: Learning to Adapt Neural Networks Using Test-Time Feedback
One For All: Video Conversation is Feasible Without Video Instruction Tuning
Convolutional Networks with Oriented 1D Kernels
Keyword: mobile
Multimodal Dataset for Localization, Mapping and Crop Monitoring in Citrus Tree Farms
Joint Computing, Pushing, and Caching Optimization for Mobile Edge Computing Networks via Soft Actor-Critic Learning
Teach and Repeat Navigation: A Robust Control Approach
GAMMA: Graspability-Aware Mobile MAnipulation Policy Learning based on Online Grasping Pose Fusion
AdaEvo: Edge-Assisted Continuous and Timely DNN Model Evolution for Mobile Devices
Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models
Neuromorphic Imaging and Classification with Graph Learning
Tracking Snake-like Robots in the Wild Using Only a Single Camera
Time-Domain Channel Measurements and Small-Scale Fading Characterization for RIS-Assisted Wireless Communication Systems
Keyword: pruning
Model-Free, Regret-Optimal Best Policy Identification in Online CMDPs
Efficient Exact Subgraph Matching via GNN-based Path Dominance Embedding (Technical Report)
CAIT: Triple-Win Compression towards High Accuracy, Fast Inference, and Favorable Transferability For ViTs
SHACIRA: Scalable HAsh-grid Compression for Implicit Neural Representations
Keyword: diffusion
PINF: Continuous Normalizing Flows for Physics-Constrained Deep Learning
Generative Residual Diffusion Modeling for Km-scale Atmospheric Downscaling
Learning Using Generated Privileged Information by Text-to-Image Diffusion Models
Maximum Diffusion Reinforcement Learning
LD4MRec: Simplifying and Powering Diffusion Model for Multimedia Recommendation
Hypergraph $p$-Laplacians, Scale Spaces, and Information Flow in Networks
DreamCom: Finetuning Text-guided Inpainting Model for Image Composition
High-Fidelity Speech Synthesis with Minimal Supervision: All Using Diffusion Models
Uncertainty Quantification via Neural Posterior Principal Components
Dynamic Prompt Learning: Addressing Cross-Attention Leakage for Text-Based Image Editing
Factorized Diffusion Architectures for Unsupervised Image Generation and Segmentation
Emu: Enhancing Image Generation Models Using Photogenic Needles in a Haystack
Show-1: Marrying Pixel and Latent Diffusion Models for Text-to-Video Generation
LGMCTS: Language-Guided Monte-Carlo Tree Search for Executable Semantic Object Rearrangement
Exploiting the Signal-Leak Bias in Diffusion Models
Keyword: adaptive
STANCE-C3: Domain-adaptive Cross-target Stance Detection via Contrastive Learning and Counterfactual Generation
Balancing Computational Efficiency and Forecast Error in Machine Learning-based Time-Series Forecasting: Insights from Live Experiments on Meteorological Nowcasting
Scaling Representation Learning from Ubiquitous ECG with State-Space Models
Seeing Beyond the Patch: Scale-Adaptive Semantic Segmentation of High-resolution Remote Sensing Imagery based on Reinforcement Learning
MINS: Efficient and Robust Multisensor-aided Inertial Navigation System
Parallel local time stepping for rigid bodies represented by triangulated meshes
Automatic Feature Fairness in Recommendation via Adversaries
NeuRBF: A Neural Fields Representation with Adaptive Radial Basis Functions
Cardinality Estimation of Subgraph Matching: A Filtering-Sampling Approach
Enabling Resource-efficient AIoT System with Cross-level Optimization: A survey
Towards Human-Like RL: Taming Non-Naturalistic Behavior in Deep RL via Adaptive Behavioral Costs in 3D Games
Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models
Uncertainty Quantification via Neural Posterior Principal Components
Keyword: quantization
Joint Computing, Pushing, and Caching Optimization for Mobile Edge Computing Networks via Soft Actor-Critic Learning
Finite Scalar Quantization: VQ-VAE Made Simple
Rethinking Channel Dimensions to Isolate Outliers for Low-bit Weight Quantization of Large Language Models
Low Latency of object detection for spikng neural network
SHACIRA: Scalable HAsh-grid Compression for Implicit Neural Representations