Abstract
Effectively medication recommendation with complex multimorbidity conditions is a critical task in healthcare. Most existing works predicted medications based on longitudinal records, which assumed the information transmitted patterns of learning longitudinal sequence data are stable and intra-visit medical events are serialized. However, the following conditions may have been ignored: 1) A more compact encoder for intra-relationship in the intra-visit medical event is urgent; 2) Strategies for learning accurate representations of the variable longitudinal sequences of patients are different. In this paper, we proposed a novel Sample-adaptive Hierarchical medicAtion Prediction nEtwork, termed SHAPE, to tackle the above challenges in the medication recommendation task. Specifically, we design a compact intra-visit set encoder to encode the relationship in the medical event for obtaining visit-level representation and then develop an inter-visit longitudinal encoder to learn the patient-level longitudinal representation efficiently. To endow the model with the capability of modeling the variable visit length, we introduce a soft curriculum learning method to assign the difficulty of each sample automatically by the visit length. Extensive experiments on a benchmark dataset verify the superiority of our model compared with several state-of-the-art baselines.
Temporal Patience: Efficient Adaptive Deep Learning for Embedded Radar Data Processing
Authors: Max Sponner, Julius Ott, Lorenzo Servadei, Bernd Waschneck, Robert Wille, Akash Kumar
Subjects: Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI); Signal Processing (eess.SP)
Abstract
Radar sensors offer power-efficient solutions for always-on smart devices, but processing the data streams on resource-constrained embedded platforms remains challenging. This paper presents novel techniques that leverage the temporal correlation present in streaming radar data to enhance the efficiency of Early Exit Neural Networks for Deep Learning inference on embedded devices. These networks add additional classifier branches between the architecture's hidden layers that allow for an early termination of the inference if their result is deemed sufficient enough by an at-runtime decision mechanism. Our methods enable more informed decisions on when to terminate the inference, reducing computational costs while maintaining a minimal loss of accuracy. Our results demonstrate that our techniques save up to 26% of operations per inference over a Single Exit Network and 12% over a confidence-based Early Exit version. Our proposed techniques work on commodity hardware and can be combined with traditional optimizations, making them accessible for resource-constrained embedded platforms commonly used in smart devices. Such efficiency gains enable real-time radar data processing on resource-constrained platforms, allowing for new applications in the context of smart homes, Internet-of-Things, and human-computer interaction.
Authors: Ivan Grishchenko, Geng Yan, Eduard Gabriel Bazavan, Andrei Zanfir, Nikolai Chinaev, Karthik Raveendran, Matthias Grundmann, Cristian Sminchisescu
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
We present Blendshapes GHUM, an on-device ML pipeline that predicts 52 facial blendshape coefficients at 30+ FPS on modern mobile phones, from a single monocular RGB image and enables facial motion capture applications like virtual avatars. Our main contributions are: i) an annotation-free offline method for obtaining blendshape coefficients from real-world human scans, ii) a lightweight real-time model that predicts blendshape coefficients based on facial landmarks.
Grey-box Bayesian Optimization for Sensor Placement in Assisted Living Environments
Authors: Shadan Golestan, Omid Ardakanian, Pierre Boulanger
Abstract
Optimizing the configuration and placement of sensors is crucial for reliable fall detection, indoor localization, and activity recognition in assisted living spaces. We propose a novel, sample-efficient approach to find a high-quality sensor placement in an arbitrary indoor space based on grey-box Bayesian optimization and simulation-based evaluation. Our key technical contribution lies in capturing domain-specific knowledge about the spatial distribution of activities and incorporating it into the iterative selection of query points in Bayesian optimization. Considering two simulated indoor environments and a real-world dataset containing human activities and sensor triggers, we show that our proposed method performs better compared to state-of-the-art black-box optimization techniques in identifying high-quality sensor placements, leading to accurate activity recognition in terms of F1-score, while also requiring a significantly lower (51.3% on average) number of expensive function queries.
PACE: Prompting and Augmentation for Calibrated Confidence Estimation with GPT-4 in Cloud Incident Root Cause Analysis
Abstract
In recent years, the transition to cloud-based platforms in the IT sector has emphasized the significance of cloud incident root cause analysis to ensure service reliability and maintain customer trust. Central to this process is the efficient determination of root causes, a task made challenging due to the complex nature of contemporary cloud infrastructures. Despite the proliferation of AI-driven tools for root cause identification, their applicability remains limited by the inconsistent quality of their outputs. This paper introduces a method for enhancing confidence estimation in root cause analysis tools by prompting retrieval-augmented large language models (LLMs). This approach operates in two phases. Initially, the model evaluates its confidence based on historical incident data, considering its assessment of the evidence strength. Subsequently, the model reviews the root cause generated by the predictor. An optimization step then combines these evaluations to determine the final confidence assignment. Experimental results illustrate that our method enables the model to articulate its confidence effectively, providing a more calibrated score. We address research questions evaluating the ability of our method to produce calibrated confidence scores using LLMs, the impact of domain-specific retrieved examples on confidence estimates, and its potential generalizability across various root cause analysis models. Through this, we aim to bridge the confidence estimation gap, aiding on-call engineers in decision-making and bolstering the efficiency of cloud incident management.
The Safety Filter: A Unified View of Safety-Critical Control in Autonomous Systems
Abstract
Recent years have seen significant progress in the realm of robot autonomy, accompanied by the expanding reach of robotic technologies. However, the emergence of new deployment domains brings unprecedented challenges in ensuring safe operation of these systems, which remains as crucial as ever. While traditional model-based safe control methods struggle with generalizability and scalability, emerging data-driven approaches tend to lack well-understood guarantees, which can result in unpredictable catastrophic failures. Successful deployment of the next generation of autonomous robots will require integrating the strengths of both paradigms. This article provides a review of safety filter approaches, highlighting important connections between existing techniques and proposing a unified technical framework to understand, compare, and combine them. The new unified view exposes a shared modular structure across a range of seemingly disparate safety filter classes and naturally suggests directions for future progress towards more scalable synthesis, robust monitoring, and efficient intervention.
Fairness- and uncertainty-aware data generation for data-driven design
Abstract
The design dataset is the backbone of data-driven design. Ideally, the dataset should be fairly distributed in both shape and property spaces to efficiently explore the underlying relationship. However, the classical experimental design focuses on shape diversity and thus yields biased exploration in the property space. Recently developed methods either conduct subset selection from a large dataset or employ assumptions with severe limitations. In this paper, fairness- and uncertainty-aware data generation (FairGen) is proposed to actively detect and generate missing properties starting from a small dataset. At each iteration, its coverage module computes the data coverage to guide the selection of the target properties. The uncertainty module ensures that the generative model can make certain and thus accurate shape predictions. Integrating the two modules, Bayesian optimization determines the target properties, which are thereafter fed into the generative model to predict the associated shapes. The new designs, whose properties are analyzed by simulation, are added to the design dataset. An S-slot design dataset case study was implemented to demonstrate the efficiency of FairGen in auxetic structural design. Compared with grid and randomized sampling, FairGen increased the coverage score at twice the speed and significantly expanded the sampled region in the property space. As a result, the generative models trained with FairGen-generated datasets showed consistent and significant reductions in mean absolute errors.
ChemSpaceAL: An Efficient Active Learning Methodology Applied to Protein-Specific Molecular Generation
Authors: Gregory W. Kyro, Anton Morgunov, Rafael I. Brent, Victor S. Batista
Abstract
The incredible capabilities of generative artificial intelligence models have inevitably led to their application in the domain of drug discovery. It is therefore of tremendous interest to develop methodologies that enhance the abilities and applicability of these powerful tools. In this work, we present a novel and efficient semi-supervised active learning methodology that allows for the fine-tuning of a generative model with respect to an objective function by strategically operating within a constructed representation of the sample space. In the context of targeted molecular generation, we demonstrate the ability to fine-tune a GPT-based molecular generator with respect to an attractive interaction-based scoring function by strategically operating within a chemical space proxy, thereby maximizing attractive interactions between the generated molecules and a protein target. Importantly, our approach does not require the individual evaluation of all data points that are used for fine-tuning, enabling the incorporation of computationally expensive metrics. We are hopeful that the inherent generality of this methodology ensures that it will remain applicable as this exciting field evolves. To facilitate implementation and reproducibility, we have made all of our software available through the open-source ChemSpaceAL Python package.
Uncovering mesa-optimization algorithms in Transformers
Authors: Johannes von Oswald, Eyvind Niklasson, Maximilian Schlegel, Seijin Kobayashi, Nicolas Zucchet, Nino Scherrer, Nolan Miller, Mark Sandler, Blaise Agüera y Arcas, Max Vladymyrov, Razvan Pascanu, João Sacramento
Abstract
Transformers have become the dominant model in deep learning, but the reason for their superior performance is poorly understood. Here, we hypothesize that the strong performance of Transformers stems from an architectural bias towards mesa-optimization, a learned process running within the forward pass of a model consisting of the following two steps: (i) the construction of an internal learning objective, and (ii) its corresponding solution found through optimization. To test this hypothesis, we reverse-engineer a series of autoregressive Transformers trained on simple sequence modeling tasks, uncovering underlying gradient-based mesa-optimization algorithms driving the generation of predictions. Moreover, we show that the learned forward-pass optimization algorithm can be immediately repurposed to solve supervised few-shot tasks, suggesting that mesa-optimization might underlie the in-context learning capabilities of large language models. Finally, we propose a novel self-attention layer, the mesa-layer, that explicitly and efficiently solves optimization problems specified in context. We find that this layer can lead to improved performance in synthetic and preliminary language modeling experiments, adding weight to our hypothesis that mesa-optimization is an important operation hidden within the weights of trained Transformers.
Force-directed graph embedding with hops distance
Authors: Hamidreza Lotfalizadeh, Mohammad Al Hasan
Abstract
Graph embedding has become an increasingly important technique for analyzing graph-structured data. By representing nodes in a graph as vectors in a low-dimensional space, graph embedding enables efficient graph processing and analysis tasks like node classification, link prediction, and visualization. In this paper, we propose a novel force-directed graph embedding method that utilizes the steady acceleration kinetic formula to embed nodes in a way that preserves graph topology and structural features. Our method simulates a set of customized attractive and repulsive forces between all node pairs with respect to their hop distance. These forces are then used in Newton's second law to obtain the acceleration of each node. The method is intuitive, parallelizable, and highly scalable. We evaluate our method on several graph analysis tasks and show that it achieves competitive performance compared to state-of-the-art unsupervised embedding techniques.
SkillScanner: Detecting Policy-Violating Voice Applications Through Static Analysis at the Development Phase
Authors: Song Liao, Long Cheng, Haipeng Cai, Linke Guo, Hongxin Hu
Abstract
The Amazon Alexa marketplace is the largest Voice Personal Assistant (VPA) platform with over 100,000 voice applications (i.e., skills) published to the skills store. In an effort to maintain the quality and trustworthiness of voice-apps, Amazon Alexa has implemented a set of policy requirements to be adhered to by third-party skill developers. However, recent works reveal the prevalence of policy-violating skills in the current skills store. To understand the causes of policy violations in skills, we first conduct a user study with 34 third-party skill developers focusing on whether they are aware of the various policy requirements defined by the Amazon Alexa platform. Our user study results show that there is a notable gap between VPA's policy requirements and skill developers' practices. As a result, it is inevitable that policy-violating skills will be published. To prevent the inflow of new policy-breaking skills to the skills store from the source, it is critical to identify potential policy violations at the development phase. In this work, we design and develop SkillScanner, an efficient static code analysis tool to facilitate third-party developers to detect policy violations early in the skill development lifecycle. To evaluate the performance of SkillScanner, we conducted an empirical study on 2,451 open source skills collected from GitHub. SkillScanner effectively identified 1,328 different policy violations from 786 skills. Our results suggest that 32% of these policy violations are introduced through code duplication (i.e., code copy and paste). In particular, we found that 42 skill code examples from potential Alexa's official accounts (e.g., "alexa" and "alexa-samples" on GitHub) contain policy violations, which lead to 81 policy violations in other skills due to the copy-pasted code snippets from these Alexa's code examples.
Systemization of Knowledge (SoK)- Cross Impact of Transfer Learning in Cybersecurity: Offensive, Defensive and Threat Intelligence Perspectives
Authors: Sofiya Makar, Ali Dehghantanha, Fattane Zarrinkalam, Gautam Srivastava, Abbas Yazdinejad
Abstract
Recent literature highlights a significant cross-impact between transfer learning and cybersecurity. Many studies have been conducted on using transfer learning to enhance security, leading to various applications in different cybersecurity tasks. However, previous research is focused on specific areas of cybersecurity. This paper presents a comprehensive survey of transfer learning applications in cybersecurity by covering a wide range of domains, identifying current trends, and shedding light on under-explored areas. The survey highlights the significance of transfer learning in addressing critical issues in cybersecurity, such as improving detection accuracy, reducing training time, handling data imbalance, and enhancing privacy preservation. Additional insights are provided on the common problems solved using transfer learning, such as the lack of labeled data, different data distributions, and privacy concerns. The paper identifies future research directions and challenges that require community attention, including the need for privacy-preserving models, automatic tools for knowledge transfer, metrics for measuring domain relatedness, and enhanced privacy preservation mechanisms. The insights and roadmap presented in this paper will guide researchers in further advancing transfer learning in cybersecurity, fostering the development of robust and efficient cybersecurity systems to counter emerging threats and protect sensitive information. To the best of our knowledge, this paper is the first of its kind to present a comprehensive taxonomy of all areas of cybersecurity that benefited from transfer learning and propose a detailed future roadmap to shape the possible research direction in this area.
Fast Constraint Screening for Multi-Interval Unit Commitment
Abstract
Power systems Unit Commitment (UC) problem determines the generator commitment schedule and dispatch decisions for power networks based on forecasted electricity demand. However, with the increasing penetration of renewables and stochastic demand behaviors, it becomes challenging to solve the large-scale, multi-interval UC problem in an efficient manner. The main objective of this paper is to propose a fast and reliable scheme to eliminate a set of redundant or inactive physical constraints in the high-dimensional, multi-interval, mixed-integer UC problem, while the reduced problem is equivalent to the original full problem in terms of commitment decisions. Our key insights lie on pre-screening the constraints based on the load distribution and considering the physical feasibility regions of multi-interval UC problem. For the multistep UC formulation, we overcome screening conservativeness by utilizing the multi-step ramping relationships, and can reliably screen out more constraints compared to current practice. Extensive simulations on both specific load samples and load regions validate the proposed technique can screen out more than 80% constraints while preserving the feasibility of multi-interval UC problem.
Geometry Enhanced Optimal Control Technique for Acrobatic Flip Motion of Quadcopter
Abstract
A nonlinear optimal control strategy, named the geometry enhanced finite time $\boldsymbol{\theta-}$D technique, is proposed to manipulate the acrobatic flip flight of variable pitch (VP) quadcopter unmanned aerial vehicles (abbreviated as VP copter). A unique superiority of the VP copter, which can provide the thrust in both positive and negative vertical directions by varying the pitch angles of blades, facilitates the acrobatic flip motion. The finite time $\boldsymbol{\theta-}$D technique can offer a closed-form near-optimal state feedback control law with online computational efficiency as compared with the finite time state-dependent Riccati equation (SDRE) technique. Meanwhile, by virtue of the geometric technique, the singularity issue of the rotation matrix in the acrobatic flip maneuver can be avoided. The simulation experiments verify the proposed control strategy is effective and efficient.
Reset Controller Synthesis by Reach-avoid Analysis for Delay Hybrid Systems
Authors: Han Su, Jiyu Zhu, Shenghua Feng, Yunjun Bai, Bin Gu, Jiang Liu, Mengfei Yang, Naijun Zhan
Abstract
A reset controller plays a crucial role in designing hybrid systems. It restricts the initial set and redefines the reset map associated with discrete transitions, in order to guarantee the system to achieve its objective. Reset controller synthesis, together with feedback controller synthesis and switching logic controller synthesis, provides a correct-by-construction approach to designing hybrid systems. However, time-delay is an inevitable factor in hybrid systems, which can degrade control performance and render verification certificates obtained by abstracting away time-delay invalid in practice. In this paper, we investigate this issue in a practical manner by taking time-delay into account. We propose an approach that reduces the synthesis of reset controllers to the generation of reach-avoid sets for the hybrid system under consideration, which can be efficiently solved using off-the-shell convex optimization solvers.
Combining deep learning and street view imagery to map smallholder crop types
Authors: Jordi Laguarta, Thomas Friedel, Sherrie Wang
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Abstract
Accurate crop type maps are an essential source of information for monitoring yield progress at scale, projecting global crop production, and planning effective policies. To date, however, crop type maps remain challenging to create in low and middle-income countries due to a lack of ground truth labels for training machine learning models. Field surveys are the gold standard in terms of accuracy but require an often-prohibitively large amount of time, money, and statistical capacity. In recent years, street-level imagery, such as Google Street View, KartaView, and Mapillary, has become available around the world. Such imagery contains rich information about crop types grown at particular locations and times. In this work, we develop an automated system to generate crop type ground references using deep learning and Google Street View imagery. The method efficiently curates a set of street view images containing crop fields, trains a model to predict crop type by utilizing weakly-labelled images from disparate out-of-domain sources, and combines predicted labels with remote sensing time series to create a wall-to-wall crop type map. We show that, in Thailand, the resulting country-wide map of rice, cassava, maize, and sugarcane achieves an accuracy of 93%. As the availability of roadside imagery expands, our pipeline provides a way to map crop types at scale around the globe, especially in underserved smallholder regions.
Random Segmentation: New Traffic Obfuscation against Packet-Size-Based Side-Channel Attacks
Authors: Mnassar Alyami, Abdulmajeed Alghamdi, Mohammed Alkhowaiter, Cliff Zou, Yan Solihin
Abstract
Despite encryption, the packet size is still visible, enabling observers to infer private information in the Internet of Things (IoT) environment (e.g., IoT device identification). Packet padding obfuscates packet-length characteristics with a high data overhead because it relies on adding noise to the data. This paper proposes a more data-efficient approach that randomizes packet sizes without adding noise. We achieve this by splitting large TCP segments into random-sized chunks; hence, the packet length distribution is obfuscated without adding noise data. Our client-server implementation using TCP sockets demonstrates the feasibility of our approach at the application level. We realize our packet size control by adjusting two local socket-programming parameters. First, we enable the TCP_NODELAY option to send out each packet with our specified length. Second, we downsize the sending buffer to prevent the sender from pushing out more data than can be received, which could disable our control of the packet sizes. We simulate our defense on a network trace of four IoT devices and show a reduction in device classification accuracy from 98% to 63%, close to random guessing. Meanwhile, the real-world data transmission experiments show that the added latency is reasonable, less than 21%, while the added packet header overhead is only about 5%.
Language Models as Black-Box Optimizers for Vision-Language Models
Authors: Samuel Yu, Shihong Liu, Zhiqiu Lin, Deepak Pathak, Deva Ramanan
Subjects: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Multimedia (cs.MM)
Abstract
Vision-language models (VLMs) pre-trained on web-scale datasets have demonstrated remarkable capabilities across a variety of vision and multimodal tasks. Currently, fine-tuning methods for VLMs mainly operate in a white-box setting, requiring access to model parameters for backpropagation. However, many VLMs rely on proprietary data and are not open-source, which restricts the use of white-box approaches for fine-tuning. Given that popular private large language models (LLMs) like ChatGPT still offer a language-based user interface, we aim to develop a novel fine-tuning approach for VLMs through natural language prompts, thereby avoiding the need to access model parameters, feature embeddings, or output logits. In this setup, we propose employing chat-based LLMs as black-box optimizers to search for the best text prompt on the illustrative task of few-shot image classification using CLIP. Specifically, we adopt an automatic "hill-climbing" procedure that converges on an effective prompt by evaluating the accuracy of current prompts and asking LLMs to refine them based on textual feedback, all within a conversational process without human-in-the-loop. In a challenging 1-shot learning setup, our simple approach surpasses the white-box continuous prompting method CoOp by an average of 1.5% across 11 datasets including ImageNet. Our approach also outperforms OpenAI's manually crafted prompts and is more efficient than other black-box methods like iterative APE. Additionally, we highlight the advantage of conversational feedback incorporating both positive and negative prompts, suggesting that LLMs can utilize the implicit "gradient" direction in textual feedback for a more efficient search. Lastly, we find that the text prompts generated through our strategy are not only more interpretable but also transfer well across different CLIP architectures in a black-box manner.
Trust-Region Neural Moving Horizon Estimation for Robots
Authors: Bingheng Wang, Xuyang Chen, Lin Zhao
Subjects: Robotics (cs.RO); Systems and Control (eess.SY)
Abstract
Accurate disturbance estimation is essential for safe robot operations. The recently proposed neural moving horizon estimation (NeuroMHE), which uses a portable neural network to model the MHE's weightings, shows promise in this context. Currently, NeuroMHE is trained through gradient descent, with its gradient computed recursively using a Kalman filter. This paper proposes a trust-region policy optimization method for training NeuroMHE. We achieve this by providing the second-order derivatives of MHE, referred to as the MHE Hessian. Remarkably, we establish that much of computation already used to obtain the gradient, especially the Kalman filter, can be efficiently reused to compute the MHE Hessian. This offers linear computational complexity relative to the MHE horizon. Through validation with an open-source real quadrotor flight dataset, our approach demonstrates data-efficient training (<5 min) and outperforms a state-of-the-art neural estimator by up to 68.1% in force estimation accuracy, utilizing only 1.4% of its network parameters. Furthermore, our method showcases enhanced robustness to network initialization compared to the gradient descent counterpart.
A correction function-based kernel-free boundary integral method for elliptic PDEs with implicitly defined interfaces
Abstract
This work addresses a novel version of the kernel-free boundary integral (KFBI) method for solving elliptic PDEs with implicitly defined irregular boundaries and interfaces. We focus on boundary value problems and interface problems, which are reformulated into boundary integral equations and solved with the matrix-free GMRES method. In the KFBI method, evaluating boundary and volume integrals only requires solving equivalent but much simpler interface problems in a bounding box, for which fast solvers such as FFTs and geometric multigrid methods are applicable. For the simple interface problem, a correction function is introduced for both the evaluation of right-hand side correction terms and the interpolation of a non-smooth potential function. A mesh-free collocation method is proposed to compute the correction function near the interface. The new method avoids complicated derivation for derivative jumps of the solution and is easy to implement, especially for the fourth-order method in three space dimensions. Various numerical examples are presented, including challenging cases such as high-contrast coefficients, arbitrarily close interfaces and heterogeneous interface problems. The reported numerical results verify that the proposed method is both accurate and efficient.
Gait Design of a Novel Arboreal Concertina Locomotion for Snake-like Robots
Abstract
In this paper, we propose a novel strategy for a snake robot to move straight up a cylindrical surface. Prior works on pole-climbing for a snake robot mainly utilized a rolling helix gait, and although proven to be efficient, it does not reassemble movements made by a natural snake. We take inspiration from nature and seek to imitate the Arboreal Concertina Locomotion (ACL) from real-life serpents. In order to represent the 3D curves that make up the key motion patterns of ACL, we establish a set of parametric equations that identify periodic functions, which produce a sequence of backbone curves. We then build up the gait equation using the curvature integration method, and finally, we propose a simple motion estimation strategy using virtual chassis and non-slip model assumptions. We present experimental results using a 20-DOF snake robot traversing outside of a straight pipe.
GVD-Exploration: An Efficient Autonomous Robot Exploration Framework Based on Fast Generalized Voronoi Diagram Extraction
Abstract
Rapidly-exploring Random Trees (RRTs) are a popular technique for autonomous exploration of mobile robots. However, the random sampling used by RRTs can result in inefficient and inaccurate frontiers extraction, which affects the exploration performance. To address the issues of slow path planning and high path cost, we propose a framework that uses a generalized Voronoi diagram (GVD) based multi-choice strategy for robot exploration. Our framework consists of three components: a novel mapping model that uses an end-to-end neural network to construct GVDs of the environments in real time; a GVD-based heuristic scheme that accelerates frontiers extraction and reduces frontiers redundancy; and a multi-choice frontiers assignment scheme that considers different types of frontiers and enables the robot to make rational decisions during the exploration process. We evaluate our method on simulation and real-world experiments and show that it outperforms RRT-based exploration methods in terms of efficiency and robustness.
Update Monte Carlo tree search (UMCTS) algorithm for heuristic global search of sizing optimization problems for truss structures
Abstract
Sizing optimization of truss structures is a complex computational problem, and the reinforcement learning (RL) is suitable for dealing with multimodal problems without gradient computations. In this paper, a new efficient optimization algorithm called update Monte Carlo tree search (UMCTS) is developed to obtain the appropriate design for truss structures. UMCTS is an RL-based method that combines the novel update process and Monte Carlo tree search (MCTS) with the upper confidence bound (UCB). Update process means that in each round, the optimal cross-sectional area of each member is determined by search tree, and its initial state is the final state in the previous round. In the UMCTS algorithm, an accelerator for the number of selections for member area and iteration number is introduced to reduce the computation time. Moreover, for each state, the average reward is replaced by the best reward collected on the simulation process to determine the optimal solution. The proposed optimization method is examined on some benchmark problems of planar and spatial trusses with discrete sizing variables to demonstrate the efficiency and validity. It is shown that the computation time for the proposed approach is at least ten times faster than the branch and bound (BB) method. The numerical results indicate that the proposed method stably achieves better solution than other conventional methods.
Real-Time Semantic Segmentation: A Brief Survey & Comparative Study in Remote Sensing
Abstract
Real-time semantic segmentation of remote sensing imagery is a challenging task that requires a trade-off between effectiveness and efficiency. It has many applications including tracking forest fires, detecting changes in land use and land cover, crop health monitoring, and so on. With the success of efficient deep learning methods (i.e., efficient deep neural networks) for real-time semantic segmentation in computer vision, researchers have adopted these efficient deep neural networks in remote sensing image analysis. This paper begins with a summary of the fundamental compression methods for designing efficient deep neural networks and provides a brief but comprehensive survey, outlining the recent developments in real-time semantic segmentation of remote sensing imagery. We examine several seminal efficient deep learning methods, placing them in a taxonomy based on the network architecture design approach. Furthermore, we evaluate the quality and efficiency of some existing efficient deep neural networks on a publicly available remote sensing semantic segmentation benchmark dataset, the OpenEarthMap. The experimental results of an extensive comparative study demonstrate that most of the existing efficient deep neural networks have good segmentation quality, but they suffer low inference speed (i.e., high latency rate), which may limit their capability of deployment in real-time applications of remote sensing image segmentation. We provide some insights into the current trend and future research directions for real-time semantic segmentation of remote sensing imagery.
A Perceptron-based Fine Approximation Technique for Linear Separation
Abstract
This paper presents a novel online learning method that aims at finding a separator hyperplane between data points labelled as either positive or negative. Since weights and biases of artificial neurons can directly be related to hyperplanes in high-dimensional spaces, the technique is applicable to train perceptron-based binary classifiers in machine learning. In case of large or imbalanced data sets, use of analytical or gradient-based solutions can become prohibitive and impractical, where heuristics and approximation techniques are still applicable. The proposed method is based on the Perceptron algorithm, however, it tunes neuron weights in just the necessary extent during searching the separator hyperplane. Due to an appropriate transformation of the initial data set we need not to consider data labels, neither the bias term. respectively, reducing separability to a one-class classification problem. The presented method has proven converge; empirical results show that it can be more efficient than the Perceptron algorithm, especially, when the size of the data set exceeds data dimensionality.
OmniSketch: Efficient Multi-Dimensional High-Velocity Stream Analytics with Arbitrary Predicates
Authors: Wieger R. Punter, Odysseas Papapetrou, Minos Garofalakis
Abstract
A key need in different disciplines is to perform analytics over fast-paced data streams, similar in nature to the traditional OLAP analytics in relational databases i.e., with filters and aggregates. Storing unbounded streams, however, is not a realistic, or desired approach due to the high storage requirements, and the delays introduced when storing massive data. Accordingly, many synopses/sketches have been proposed that can summarize the stream in small memory (usually sufficiently small to be stored in RAM), such that aggregate queries can be efficiently approximated, without storing the full stream. However, past synopses predominantly focus on summarizing single-attribute streams, and cannot handle filters and constraints on arbitrary subsets of multiple attributes efficiently. In this work, we propose OmniSketch, the first sketch that scales to fast-paced and complex data streams (with many attributes), and supports aggregates with filters on multiple attributes, dynamically chosen at query time. The sketch offers probabilistic guarantees, a favorable space-accuracy tradeoff, and a worst-case logarithmic complexity for updating and for query execution. We demonstrate experimentally with both real and synthetic data that the sketch outperforms the state-of-the-art, and that it can approximate complex ad-hoc queries within the configured accuracy guarantees, with small memory requirements.
Information Flow in Graph Neural Networks: A Clinical Triage Use Case
Abstract
Graph Neural Networks (GNNs) have gained popularity in healthcare and other domains due to their ability to process multi-modal and multi-relational graphs. However, efficient training of GNNs remains challenging, with several open research questions. In this paper, we investigate how the flow of embedding information within GNNs affects the prediction of links in Knowledge Graphs (KGs). Specifically, we propose a mathematical model that decouples the GNN connectivity from the connectivity of the graph data and evaluate the performance of GNNs in a clinical triage use case. Our results demonstrate that incorporating domain knowledge into the GNN connectivity leads to better performance than using the same connectivity as the KG or allowing unconstrained embedding propagation. Moreover, we show that negative edges play a crucial role in achieving good predictions, and that using too many GNN layers can degrade performance.
Measuring Catastrophic Forgetting in Cross-Lingual Transfer Paradigms: Exploring Tuning Strategies
Authors: Boshko Koloski, Blaž Škrlj, Marko Robnik-Šikonja, Senja Pollak
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Abstract
The cross-lingual transfer is a promising technique to solve tasks in less-resourced languages. In this empirical study, we compare two fine-tuning approaches combined with zero-shot and full-shot learning approaches for large language models in a cross-lingual setting. As fine-tuning strategies, we compare parameter-efficient adapter methods with fine-tuning of all parameters. As cross-lingual transfer strategies, we compare the intermediate-training (\textit{IT}) that uses each language sequentially and cross-lingual validation (\textit{CLV}) that uses a target language already in the validation phase of fine-tuning. We assess the success of transfer and the extent of catastrophic forgetting in a source language due to cross-lingual transfer, i.e., how much previously acquired knowledge is lost when we learn new information in a different language. The results on two different classification problems, hate speech detection and product reviews, each containing datasets in several languages, show that the \textit{IT} cross-lingual strategy outperforms \textit{CLV} for the target language. Our findings indicate that, in the majority of cases, the \textit{CLV} strategy demonstrates superior retention of knowledge in the base language (English) compared to the \textit{IT} strategy, when evaluating catastrophic forgetting in multiple cross-lingual transfers.
HOC-Search: Efficient CAD Model and Pose Retrieval from RGB-D Scans
Authors: Stefan Ainetter, Sinisa Stekovic, Friedrich Fraundorfer, Vincent Lepetit
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
We present an automated and efficient approach for retrieving high-quality CAD models of objects and their poses in a scene captured by a moving RGB-D camera. We first investigate various objective functions to measure similarity between a candidate CAD object model and the available data, and the best objective function appears to be a "render-and-compare" method comparing depth and mask rendering. We thus introduce a fast-search method that approximates an exhaustive search based on this objective function for simultaneously retrieving the object category, a CAD model, and the pose of an object given an approximate 3D bounding box. This method involves a search tree that organizes the CAD models and object properties including object category and pose for fast retrieval and an algorithm inspired by Monte Carlo Tree Search, that efficiently searches this tree. We show that this method retrieves CAD models that fit the real objects very well, with a speed-up factor of 10x to 120x compared to exhaustive search.
Inspection planning under execution uncertainty
Authors: Shmuel David Alpert, Kiril Solovey, Itzik Klein, Oren Salzman
Abstract
Autonomous inspection tasks necessitate effective path-planning mechanisms to efficiently gather observations from points of interest (POI). However, localization errors commonly encountered in urban environments can introduce execution uncertainty, posing challenges to the successful completion of such tasks. To tackle these challenges, we present IRIS-under uncertainty (IRIS-U^2), an extension of the incremental random inspection-roadmap search (IRIS) algorithm, that addresses the offline planning problem via an A*-based approach, where the planning process occurs prior the online execution. The key insight behind IRIS-U^2 is transforming the computed localization uncertainty, obtained through Monte Carlo (MC) sampling, into a POI probability. IRIS-U^2 offers insights into the expected performance of the execution task by providing confidence intervals (CI) for the expected coverage, expected path length, and collision probability, which becomes progressively tighter as the number of MC samples increase. The efficacy of IRIS-U^2 is demonstrated through a case study focusing on structural inspections of bridges. Our approach exhibits improved expected coverage, reduced collision probability, and yields increasingly-precise CIs as the number of MC samples grows. Furthermore, we emphasize the potential advantages of computing bounded sub-optimal solutions to reduce computation time while still maintaining the same CI boundaries.
Dynamic Visual Prompt Tuning for Parameter Efficient Transfer Learning
Authors: Chunqing Ruan, Hongjian Wang
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Parameter efficient transfer learning (PETL) is an emerging research spot that aims to adapt large-scale pre-trained models to downstream tasks. Recent advances have achieved great success in saving storage and computation costs. However, these methods do not take into account instance-specific visual clues for visual tasks. In this paper, we propose a Dynamic Visual Prompt Tuning framework (DVPT), which can generate a dynamic instance-wise token for each image. In this way, it can capture the unique visual feature of each image, which can be more suitable for downstream visual tasks. We designed a Meta-Net module that can generate learnable prompts based on each image, thereby capturing dynamic instance-wise visual features. Extensive experiments on a wide range of downstream recognition tasks show that DVPT achieves superior performance than other PETL methods. More importantly, DVPT even outperforms full fine-tuning on 17 out of 19 downstream tasks while maintaining high parameter efficiency. Our code will be released soon.
Accelerating Edge AI with Morpher: An Integrated Design, Compilation and Simulation Framework for CGRAs
Abstract
Coarse-Grained Reconfigurable Arrays (CGRAs) hold great promise as power-efficient edge accelerator, offering versatility beyond AI applications. Morpher, an open-source, architecture-adaptive CGRA design framework, is specifically designed to explore the vast design space of CGRAs. The comprehensive ecosystem of Morpher includes a tailored compiler, simulator, accelerator synthesis, and validation framework. This study provides an overview of Morpher, highlighting its capabilities in automatically compiling AI application kernels onto user-defined CGRA architectures and verifying their functionality. Through the Morpher framework, the versatility of CGRAs is harnessed to facilitate efficient compilation and verification of edge AI applications, covering important kernels representative of a wide range of embedded AI workloads. Morpher is available online at https://github.com/ecolab-nus/morpher-v2.
Active Label Refinement for Semantic Segmentation of Satellite Images
Authors: Tuan Pham Minh, Jayan Wijesingha, Daniel Kottke, Marek Herde, Denis Huseljic, Bernhard Sick, Michael Wachendorf, Thomas Esch
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Remote sensing through semantic segmentation of satellite images contributes to the understanding and utilisation of the earth's surface. For this purpose, semantic segmentation networks are typically trained on large sets of labelled satellite images. However, obtaining expert labels for these images is costly. Therefore, we propose to rely on a low-cost approach, e.g. crowdsourcing or pretrained networks, to label the images in the first step. Since these initial labels are partially erroneous, we use active learning strategies to cost-efficiently refine the labels in the second step. We evaluate the active learning strategies using satellite images of Bengaluru in India, labelled with land cover and land use labels. Our experimental results suggest that an active label refinement to improve the semantic segmentation network's performance is beneficial.
Integration of Quantum Accelerators with High Performance Computing $\unicode{x2013}$ A Review of Quantum Programming Tools
Authors: Amr Elsharkawy, Xiao-Ting Michelle To, Philipp Seitz, Yanbin Chen, Yannick Stade, Manuel Geiger, Qunsheng Huang, Xiaorang Guo, Muhammad Arslan Ansari, Christian B. Mendl, Dieter Kranzlmüller, Martin Schulz
Abstract
Quantum computing (QC) introduces a novel mode of computation with the possibility of greater computational power that remains to be exploited $\unicode{x2013}$ presenting exciting opportunities for high performance computing (HPC) applications. However, recent advancements in the field have made clear that QC does not supplant conventional HPC, but can rather be incorporated into current heterogeneous HPC infrastructures as an additional accelerator, thereby enabling the optimal utilization of both paradigms. The desire for such integration significantly affects the development of software for quantum computers, which in turn influences the necessary software infrastructure. To date, previous review papers have investigated various quantum programming tools (QPTs) (such as languages, libraries, frameworks) in their ability to program, compile, and execute quantum circuits. However, the integration effort with classical HPC frameworks or systems has not been addressed. This study aims to characterize existing QPTs from an HPC perspective, investigating if existing QPTs have the potential to be efficiently integrated with classical computing models and determining where work is still required. This work structures a set of criteria into an analysis blueprint that enables HPC scientists to assess whether a QPT is suitable for the quantum-accelerated classical application at hand.
AKEM: Aligning Knowledge Base to Queries with Ensemble Model for Entity Recognition and Linking
Authors: Di Lu, Zhongping Liang, Caixia Yuan, Xiaojie Wang
Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR)
Abstract
This paper presents a novel approach to address the Entity Recognition and Linking Challenge at NLPCC 2015. The task involves extracting named entity mentions from short search queries and linking them to entities within a reference Chinese knowledge base. To tackle this problem, we first expand the existing knowledge base and utilize external knowledge to identify candidate entities, thereby improving the recall rate. Next, we extract features from the candidate entities and utilize Support Vector Regression and Multiple Additive Regression Tree as scoring functions to filter the results. Additionally, we apply rules to further refine the results and enhance precision. Our method is computationally efficient and achieves an F1 score of 0.535.
Efficient Memory Management for Large Language Model Serving with PagedAttention
Authors: Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph E. Gonzalez, Hao Zhang, Ion Stoica
Subjects: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
Abstract
High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time. However, existing systems struggle because the key-value cache (KV cache) memory for each request is huge and grows and shrinks dynamically. When managed inefficiently, this memory can be significantly wasted by fragmentation and redundant duplication, limiting the batch size. To address this problem, we propose PagedAttention, an attention algorithm inspired by the classical virtual memory and paging techniques in operating systems. On top of it, we build vLLM, an LLM serving system that achieves (1) near-zero waste in KV cache memory and (2) flexible sharing of KV cache within and across requests to further reduce memory usage. Our evaluations show that vLLM improves the throughput of popular LLMs by 2-4$\times$ with the same level of latency compared to the state-of-the-art systems, such as FasterTransformer and Orca. The improvement is more pronounced with longer sequences, larger models, and more complex decoding algorithms. vLLM's source code is publicly available at https://github.com/vllm-project/vllm
360$^\circ$ from a Single Camera: A Few-Shot Approach for LiDAR Segmentation
Authors: Laurenz Reichardt, Nikolas Ebert, Oliver Wasenmüller
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Abstract
Deep learning applications on LiDAR data suffer from a strong domain gap when applied to different sensors or tasks. In order for these methods to obtain similar accuracy on different data in comparison to values reported on public benchmarks, a large scale annotated dataset is necessary. However, in practical applications labeled data is costly and time consuming to obtain. Such factors have triggered various research in label-efficient methods, but a large gap remains to their fully-supervised counterparts. Thus, we propose ImageTo360, an effective and streamlined few-shot approach to label-efficient LiDAR segmentation. Our method utilizes an image teacher network to generate semantic predictions for LiDAR data within a single camera view. The teacher is used to pretrain the LiDAR segmentation student network, prior to optional fine-tuning on 360$^\circ$ data. Our method is implemented in a modular manner on the point level and as such is generalizable to different architectures. We improve over the current state-of-the-art results for label-efficient methods and even surpass some traditional fully-supervised segmentation networks.
High Order Numerical Methods To Approximate The Singular Value Decomposition
Abstract
In this paper, we present a class of high order methods to approximate the singular value decomposition of a given complex matrix (SVD). To the best of our knowledge, only methods up to order three appear in the the literature. A first part is dedicated to defline and analyse this class of method in the regular case, i.e., when the singular values are pairwise distinct. The construction is based on a perturbation analysis of a suitable system of associated to the SVD (SVD system). More precisely, for an integer $p$ be given, we define a sequence which converges with an order $p + 1$ towards the left-right singular vectors and the singular values if the initial approximation of the SVD system satisfies a condition which depends on three quantities : the norm of initial approximation of the SVD system, the greatest singular value and the greatest inverse of the modulus of the difference between the singular values. From a numerical computational point of view, this furnishes a very efficient simple test to prove and certifiy the existence of a SVD in neighborhood of the initial approximation. We generalize these result in the case of clusters of singular values. We show also how to use the result of regular case to detect the clusters of singular values and to define a notion of deflation of the SVD. Moreover numerical experiments confirm the theoretical results.
SGFeat: Salient Geometric Feature for Point Cloud Registration
Authors: Qianliang Wu, Yaqing Ding, Lei Luo, Chuanwei Zhou, Jin Xie, Jian Yang
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Point Cloud Registration (PCR) is a critical and challenging task in computer vision. One of the primary difficulties in PCR is identifying salient and meaningful points that exhibit consistent semantic and geometric properties across different scans. Previous methods have encountered challenges with ambiguous matching due to the similarity among patch blocks throughout the entire point cloud and the lack of consideration for efficient global geometric consistency. To address these issues, we propose a new framework that includes several novel techniques. Firstly, we introduce a semantic-aware geometric encoder that combines object-level and patch-level semantic information. This encoder significantly improves registration recall by reducing ambiguity in patch-level superpoint matching. Additionally, we incorporate a prior knowledge approach that utilizes an intrinsic shape signature to identify salient points. This enables us to extract the most salient super points and meaningful dense points in the scene. Secondly, we introduce an innovative transformer that encodes High-Order (HO) geometric features. These features are crucial for identifying salient points within initial overlap regions while considering global high-order geometric consistency. To optimize this high-order transformer further, we introduce an anchor node selection strategy. By encoding inter-frame triangle or polyhedron consistency features based on these anchor nodes, we can effectively learn high-order geometric features of salient super points. These high-order features are then propagated to dense points and utilized by a Sinkhorn matching module to identify key correspondences for successful registration. In our experiments conducted on well-known datasets such as 3DMatch/3DLoMatch and KITTI, our approach has shown promising results, highlighting the effectiveness of our novel method.
Abstract
A Stochastic Control Problem can be solved by Dynamic Programming or Distributed Optimal Control with the Kolmogorov equation for the probability density of the Markov process of the problem. It can be solved also with Supervised Learning. We shall compare these two classes of methods for the control of fisheries. Fishing quotas are unpleasant but efficient to control the productivity of a fishing site. A popular model has a vector-valued stochastic differential equation for the biomass of the different species. Optimization of quota will be obtained by a gradient method applied to the least square difference with an ideal state weighted by the probability density of the biomasses. Alternatively a deep neural network which preserves the Markov property of the problem can be trained with a stochastic gradient algorithm. The model is extended to distributed fishing sites and biomass is stabilized by adjusting the quota to its time derivative.
Speciality vs Generality: An Empirical Study on Catastrophic Forgetting in Fine-tuning Foundation Models
Abstract
Foundation models, including Vision Language Models (VLMs) and Large Language Models (LLMs), possess the $generality$ to handle diverse distributions and tasks, which stems from their extensive pre-training datasets. The fine-tuning of foundation models is a common practice to enhance task performance or align the model's behavior with human expectations, allowing them to gain $speciality$. However, the small datasets used for fine-tuning may not adequately cover the diverse distributions and tasks encountered during pre-training. Consequently, the pursuit of speciality during fine-tuning can lead to a loss of {generality} in the model, which is related to catastrophic forgetting (CF) in deep learning. In this study, we demonstrate this phenomenon in both VLMs and LLMs. For instance, fine-tuning VLMs like CLIP on ImageNet results in a loss of generality in handling diverse distributions, and fine-tuning LLMs like Galactica in the medical domain leads to a loss in following instructions and common sense. To address the trade-off between the speciality and generality, we investigate multiple regularization methods from continual learning, the weight averaging method (Wise-FT) from out-of-distributional (OOD) generalization, which interpolates parameters between pre-trained and fine-tuned models, and parameter-efficient fine-tuning methods like Low-Rank Adaptation (LoRA). Our findings show that both continual learning and Wise-ft methods effectively mitigate the loss of generality, with Wise-FT exhibiting the strongest performance in balancing speciality and generality.
OTAS: Unsupervised Boundary Detection for Object-Centric Temporal Action Segmentation
Authors: Yuerong Li, Zhengrong Xue, Huazhe Xu
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Temporal action segmentation is typically achieved by discovering the dramatic variances in global visual descriptors. In this paper, we explore the merits of local features by proposing the unsupervised framework of Object-centric Temporal Action Segmentation (OTAS). Broadly speaking, OTAS consists of self-supervised global and local feature extraction modules as well as a boundary selection module that fuses the features and detects salient boundaries for action segmentation. As a second contribution, we discuss the pros and cons of existing frame-level and boundary-level evaluation metrics. Through extensive experiments, we find OTAS is superior to the previous state-of-the-art method by $41\%$ on average in terms of our recommended F1 score. Surprisingly, OTAS even outperforms the ground-truth human annotations in the user study. Moreover, OTAS is efficient enough to allow real-time inference.
Transferability analysis of data-driven additive manufacturing knowledge: a case study between powder bed fusion and directed energy deposition
Authors: Mutahar Safdar, Jiarui Xie, Hyunwoong Ko, Yan Lu, Guy Lamouche, Yaoyao Fiona Zhao
Abstract
Data-driven research in Additive Manufacturing (AM) has gained significant success in recent years. This has led to a plethora of scientific literature to emerge. The knowledge in these works consists of AM and Artificial Intelligence (AI) contexts that have not been mined and formalized in an integrated way. Moreover, no tools or guidelines exist to support data-driven knowledge transfer from one context to another. As a result, data-driven solutions using specific AI techniques are being developed and validated only for specific AM process technologies. There is a potential to exploit the inherent similarities across various AM technologies and adapt the existing solutions from one process or problem to another using AI, such as Transfer Learning. We propose a three-step knowledge transferability analysis framework in AM to support data-driven AM knowledge transfer. As a prerequisite to transferability analysis, AM knowledge is featurized into identified knowledge components. The framework consists of pre-transfer, transfer, and post-transfer steps to accomplish knowledge transfer. A case study is conducted between flagship metal AM processes. Laser Powder Bed Fusion (LPBF) is the source of knowledge motivated by its relative matureness in applying AI over Directed Energy Deposition (DED), which drives the need for knowledge transfer as the less explored target process. We show successful transfer at different levels of the data-driven solution, including data representation, model architecture, and model parameters. The pipeline of AM knowledge transfer can be automated in the future to allow efficient cross-context or cross-process knowledge exchange.
Using conservative voltage reduction and dynamic thermal rating for congestion management of power network
Abstract
Increasing the amount of electric power that is used on the demand side has brought more attention to the peak-load management of the distribution network (DN). The creation of infrastructures for smart grids, the efficient utilization of the distributed network's components, and the appropriate administration of the distributed network would result in a valuable solution for the operators of the distributed network. As a result, a framework for peak-load management is given in this research. Within this framework, the real-time rating of the components and the voltage-dependent characteristics of the electric loads work together to assist the DN operator in effectively navigating peak periods. The combination of the conservation voltage reduction (CVR) and the dynamic thermal rating (DTR) of the components that make up the DN produces outcomes that are more helpful than any of these factors alone could provide. This is true even though each of these factors contributes to the efficient functioning of the DN. According to the findings, as compared to the individual implementation of CVR, the simultaneous utilization of DTR and CVR results in a cost-savings rise at peak events which is 58.75 percentage points more than the individual implementation. In addition, a discussion is offered concerning the current difficulties that are being experienced by the feeders that are providing the voltage-dependent constant-power loads during the utilization of the CVR, which are handled by the dynamic rating of the components that make up the DN.
Distributed Precoding for Satellite-Terrestrial Integrated Networks Without Sharing CSIT: A Rate-Splitting Approach
Authors: Doseon Kim, Wonjae Shin, Jeonghun Park, Dong Ku Kim
Abstract
Satellite-terrestrial integrated networks (STINs) are promising architecture for providing global coverage. In STINs, full frequency reuse between a satellite and a terrestrial base station (BS) is encouraged for enhancing spectral efficiency, which accounts for non-negligible amount of interference. To address the interference management problem in STINs, this paper proposes a novel distributed precoding method. Key features of our method are: i) a rate-splitting (RS) strategy is incorporated for efficient interference management, ii) precoders are designed in a distributed way without sharing channel state information between a satellite and a terrestrial BS. Specifically, to design precoders in a distributed fashion, we put forth a spectral efficiency decoupling technique. This technique disentangles the total spectral efficiency into two distinct terms, each dependent solely on the satellite's precoder and the terrestrial BS's precoder, respectively. Then, to resolve the non-smoothness raised by adopting the RS strategy, we approximate the spectral efficiency expression as a smooth function; thereafter we develop a generalized power iteration inspired optimization algorithm built based on the first-order optimality condition. Simulation results demonstrate that the proposed method improves the spectral efficiency (around 20~29%) compared to existing distributed precoding schemes.
A Simple Multiple-Access Design for Reconfigurable Intelligent Surface-Aided Systems
Authors: Wei Jiang, Hans D. Schotten
Subjects: Information Theory (cs.IT); Signal Processing (eess.SP)
Abstract
This paper focuses on the design of transmission methods and reflection optimization for a wireless system assisted by a single or multiple reconfigurable intelligent surfaces (RISs). The existing techniques are either too complex to implement in practical systems or too inefficient to achieve high performance. To overcome the shortcomings of the existing schemes, we propose a simple but efficient approach based on \textit{opportunistic reflection} and \textit{non-orthogonal transmission}. The key idea is opportunistically selecting the best user that can reap the maximal gain from the optimally reflected signals via RIS. That is to say, only the channel state information of the best user is used for RIS reflection optimization, which can in turn lower complexity substantially. In addition, the second user is selected to superpose its signal on that of the primary user, where the benefits of non-orthogonal transmission, i.e., high system capacity and improved user fairness, are obtained. Additionally, a simplified variant exploiting random phase shifts is proposed to avoid the high overhead of RIS channel estimation.
Enhancing In-Memory Spatial Indexing with Learned Search
Authors: Varun Pandey, Alexander van Renen, Eleni Tzirita Zacharatou, Andreas Kipf, Ibrahim Sabek, Jialin Ding, Volker Markl, Alfons Kemper
Abstract
Spatial data is ubiquitous. Massive amounts of data are generated every day from a plethora of sources such as billions of GPS-enabled devices (e.g., cell phones, cars, and sensors), consumer-based applications (e.g., Uber and Strava), and social media platforms (e.g., location-tagged posts on Facebook, Twitter, and Instagram). This exponential growth in spatial data has led the research community to build systems and applications for efficient spatial data processing. In this study, we apply a recently developed machine-learned search technique for single-dimensional sorted data to spatial indexing. Specifically, we partition spatial data using six traditional spatial partitioning techniques and employ machine-learned search within each partition to support point, range, distance, and spatial join queries. Adhering to the latest research trends, we tune the partitioning techniques to be instance-optimized. By tuning each partitioning technique for optimal performance, we demonstrate that: (i) grid-based index structures outperform tree-based index structures (from 1.23$\times$ to 2.47$\times$), (ii) learning-enhanced variants of commonly used spatial index structures outperform their original counterparts (from 1.44$\times$ to 53.34$\times$ faster), (iii) machine-learned search within a partition is faster than binary search by 11.79% - 39.51% when filtering on one dimension, (iv) the benefit of machine-learned search diminishes in the presence of other compute-intensive operations (e.g. scan costs in higher selectivity queries, Haversine distance computation, and point-in-polygon tests), and (v) index lookup is the bottleneck for tree-based structures, which could potentially be reduced by linearizing the indexed partitions.
Human-Centered Autonomy for Autonomous sUAS Target Searching
Authors: Hunter M. Ray, Zakariya Laouar, Zachary Sunberg, Nisar Ahmed
Abstract
Deploying robots that operate in dynamic, uncertain environments, such as Uncrewed Aerial Systems in search \& rescue missions, require nearly continuous human supervision for vehicle guidance and operation. Without approaches that consider high level mission context, operational methods of autonomous flying necessitate cumbersome manual operation or inefficient exhaustive search patterns. To facilitate more effective use of autonomy, we present a human-centered autonomous system that infers geospatial mission context through dynamic features sets, which then guides a probabilistic target search planner. Operators provide a limited set of diverse inputs, including priority definition, spatial semantic observations over ad-hoc geographical areas, and reference waypoints, which are probabilistically fused with geographical database information and condensed into a discretized value map representing an operator's preferences over an operational area. An online, POMDP-based planner, optimized for target searching, is augmented with this value map to generate an operator-constrained vehicle waypoint guidance plan. We validate the system by gathering input from five first responders trained in search \& rescue and compare simulated system performance against current operational methods for autonomous missions. These results display effective task mental model alignment and more efficient guidance plans, resulting in faster rescue times.
On Computationally Efficient Learning of Exponential Family Distributions
Authors: Abhin Shah, Devavrat Shah, Gregory W. Wornell
Abstract
We consider the classical problem of learning, with arbitrary accuracy, the natural parameters of a $k$-parameter truncated \textit{minimal} exponential family from i.i.d. samples in a computationally and statistically efficient manner. We focus on the setting where the support as well as the natural parameters are appropriately bounded. While the traditional maximum likelihood estimator for this class of exponential family is consistent, asymptotically normal, and asymptotically efficient, evaluating it is computationally hard. In this work, we propose a novel loss function and a computationally efficient estimator that is consistent as well as asymptotically normal under mild conditions. We show that, at the population level, our method can be viewed as the maximum likelihood estimation of a re-parameterized distribution belonging to the same class of exponential family. Further, we show that our estimator can be interpreted as a solution to minimizing a particular Bregman score as well as an instance of minimizing the \textit{surrogate} likelihood. We also provide finite sample guarantees to achieve an error (in $\ell_2$-norm) of $\alpha$ in the parameter estimation with sample complexity $O({\sf poly}(k)/\alpha^2)$. Our method achives the order-optimal sample complexity of $O({\sf log}(k)/\alpha^2)$ when tailored for node-wise-sparse Markov random fields. Finally, we demonstrate the performance of our estimator via numerical experiments.
C4CAM: A Compiler for CAM-based In-memory Accelerators
Authors: Hamid Farzaneh, João Paulo Cardoso de Lima, Mengyuan Li, Asif Ali Khan, Xiaobo Sharon Hu, Jeronimo Castrillon
Abstract
Machine learning and data analytics applications increasingly suffer from the high latency and energy consumption of conventional von Neumann architectures. Recently, several in-memory and near-memory systems have been proposed to remove this von Neumann bottleneck. Platforms based on content-addressable memories (CAMs) are particularly interesting due to their efficient support for the search-based operations that form the foundation for many applications, including K-nearest neighbors (KNN), high-dimensional computing (HDC), recommender systems, and one-shot learning among others. Today, these platforms are designed by hand and can only be programmed with low-level code, accessible only to hardware experts. In this paper, we introduce C4CAM, the first compiler framework to quickly explore CAM configurations and to seamlessly generate code from high-level TorchScript code. C4CAM employs a hierarchy of abstractions that progressively lowers programs, allowing code transformations at the most suitable abstraction level. Depending on the type and technology, CAM arrays exhibit varying latencies and power profiles. Our framework allows analyzing the impact of such differences in terms of system-level performance and energy consumption, and thus supports designers in selecting appropriate designs for a given application.
Keyword: faster
Mobile Vision Transformer-based Visual Object Tracking
Authors: Goutam Yelluru Gopal, Maria A. Amer
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
The introduction of robust backbones, such as Vision Transformers, has improved the performance of object tracking algorithms in recent years. However, these state-of-the-art trackers are computationally expensive since they have a large number of model parameters and rely on specialized hardware (e.g., GPU) for faster inference. On the other hand, recent lightweight trackers are fast but are less accurate, especially on large-scale datasets. We propose a lightweight, accurate, and fast tracking algorithm using Mobile Vision Transformers (MobileViT) as the backbone for the first time. We also present a novel approach of fusing the template and search region representations in the MobileViT backbone, thereby generating superior feature encoding for target localization. The experimental results show that our MobileViT-based Tracker, MVT, surpasses the performance of recent lightweight trackers on the large-scale datasets GOT10k and TrackingNet, and with a high inference speed. In addition, our method outperforms the popular DiMP-50 tracker despite having 4.7 times fewer model parameters and running at 2.8 times its speed on a GPU. The tracker code and models are available at https://github.com/goutamyg/MVT
Analysis of Randomized Householder-Cholesky QR Factorization with Multisketching
Authors: Andrew J. Higgins, Daniel B. Szyld, Erik G. Boman, Ichitaro Yamazaki
Abstract
CholeskyQR2 and shifted CholeskyQR3 are two state-of-the-art algorithms for computing tall-and-skinny QR factorizations since they attain high performance on current computer architectures. However, to guarantee stability, for some applications, CholeskyQR2 faces a prohibitive restriction on the condition number of the underlying matrix to factorize. Shifted CholeskyQR3 is stable but has $50\%$ more computational and communication costs than CholeskyQR2. In this paper, a randomized QR algorithm called Randomized Householder-Cholesky (\texttt{rand_cholQR}) is proposed and analyzed. Using one or two random sketch matrices, it is proved that with high probability, its orthogonality error is bounded by a constant of the order of unit roundoff for any numerically full-rank matrix, and hence it is as stable as shifted CholeskyQR3. An evaluation of the performance of \texttt{rand_cholQR} on a NVIDIA A100 GPU demonstrates that for tall-and-skinny matrices, \texttt{rand_cholQR} with multiple sketch matrices is nearly as fast as, or in some cases faster than, CholeskyQR2. Hence, compared to CholeskyQR2, \texttt{rand_cholQR} is more stable with almost no extra computational or memory cost, and therefore a superior algorithm both in theory and practice.
Update Monte Carlo tree search (UMCTS) algorithm for heuristic global search of sizing optimization problems for truss structures
Abstract
Sizing optimization of truss structures is a complex computational problem, and the reinforcement learning (RL) is suitable for dealing with multimodal problems without gradient computations. In this paper, a new efficient optimization algorithm called update Monte Carlo tree search (UMCTS) is developed to obtain the appropriate design for truss structures. UMCTS is an RL-based method that combines the novel update process and Monte Carlo tree search (MCTS) with the upper confidence bound (UCB). Update process means that in each round, the optimal cross-sectional area of each member is determined by search tree, and its initial state is the final state in the previous round. In the UMCTS algorithm, an accelerator for the number of selections for member area and iteration number is introduced to reduce the computation time. Moreover, for each state, the average reward is replaced by the best reward collected on the simulation process to determine the optimal solution. The proposed optimization method is examined on some benchmark problems of planar and spatial trusses with discrete sizing variables to demonstrate the efficiency and validity. It is shown that the computation time for the proposed approach is at least ten times faster than the branch and bound (BB) method. The numerical results indicate that the proposed method stably achieves better solution than other conventional methods.
Efficient Memory Management for Large Language Model Serving with PagedAttention
Authors: Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph E. Gonzalez, Hao Zhang, Ion Stoica
Subjects: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
Abstract
High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time. However, existing systems struggle because the key-value cache (KV cache) memory for each request is huge and grows and shrinks dynamically. When managed inefficiently, this memory can be significantly wasted by fragmentation and redundant duplication, limiting the batch size. To address this problem, we propose PagedAttention, an attention algorithm inspired by the classical virtual memory and paging techniques in operating systems. On top of it, we build vLLM, an LLM serving system that achieves (1) near-zero waste in KV cache memory and (2) flexible sharing of KV cache within and across requests to further reduce memory usage. Our evaluations show that vLLM improves the throughput of popular LLMs by 2-4$\times$ with the same level of latency compared to the state-of-the-art systems, such as FasterTransformer and Orca. The improvement is more pronounced with longer sequences, larger models, and more complex decoding algorithms. vLLM's source code is publicly available at https://github.com/vllm-project/vllm
The Time Complexity of Fully Sparse Matrix Multiplication
Authors: Amir Abboud, Karl Bringmann, Nick Fischer, Marvin Künnemann
Abstract
What is the time complexity of matrix multiplication of sparse integer matrices with $m{in}$ nonzeros in the input and $m{out}$ nonzeros in the output? This paper provides improved upper bounds for this question for almost any choice of $m{in}$ vs. $m{out}$, and provides evidence that these new bounds might be optimal up to further progress on fast matrix multiplication. Our main contribution is a new algorithm that reduces sparse matrix multiplication to dense (but smaller) rectangular matrix multiplication. Our running time thus depends on the optimal exponent $\omega(a,b,c)$ of multiplying dense $n^a\times n^b$ by $n^b\times n^c$ matrices. We discover that when $m{out}=\Theta(m{in}^r)$ the time complexity of sparse matrix multiplication is $O(m{in}^{\sigma+\epsilon})$, for all $\epsilon > 0$, where $\sigma$ is the solution to the equation $\omega(\sigma-1,2-\sigma,1+r-\sigma)=\sigma$. No matter what $\omega(\cdot,\cdot,\cdot)$ turns out to be, and for all $r\in(0,2)$, the new bound beats the state of the art, and we provide evidence that it is optimal based on the complexity of the all-edge triangle problem. In particular, in terms of the input plus output size $m = m{in} + m_{out}$ our algorithm runs in time $O(m^{1.3459})$. Even for Boolean matrices, this improves over the previous $m^{\frac{2\omega}{\omega+1}+\epsilon}=O(m^{1.4071})$ bound [Amossen, Pagh; 2009], which was a natural barrier since it coincides with the longstanding bound of all-edge triangle in sparse graphs [Alon, Yuster, Zwick; 1994]. We find it interesting that matrix multiplication can be solved faster than triangle detection in this natural setting. In fact, we establish an equivalence to a special case of the all-edge triangle problem.
Enhancing In-Memory Spatial Indexing with Learned Search
Authors: Varun Pandey, Alexander van Renen, Eleni Tzirita Zacharatou, Andreas Kipf, Ibrahim Sabek, Jialin Ding, Volker Markl, Alfons Kemper
Abstract
Spatial data is ubiquitous. Massive amounts of data are generated every day from a plethora of sources such as billions of GPS-enabled devices (e.g., cell phones, cars, and sensors), consumer-based applications (e.g., Uber and Strava), and social media platforms (e.g., location-tagged posts on Facebook, Twitter, and Instagram). This exponential growth in spatial data has led the research community to build systems and applications for efficient spatial data processing. In this study, we apply a recently developed machine-learned search technique for single-dimensional sorted data to spatial indexing. Specifically, we partition spatial data using six traditional spatial partitioning techniques and employ machine-learned search within each partition to support point, range, distance, and spatial join queries. Adhering to the latest research trends, we tune the partitioning techniques to be instance-optimized. By tuning each partitioning technique for optimal performance, we demonstrate that: (i) grid-based index structures outperform tree-based index structures (from 1.23$\times$ to 2.47$\times$), (ii) learning-enhanced variants of commonly used spatial index structures outperform their original counterparts (from 1.44$\times$ to 53.34$\times$ faster), (iii) machine-learned search within a partition is faster than binary search by 11.79% - 39.51% when filtering on one dimension, (iv) the benefit of machine-learned search diminishes in the presence of other compute-intensive operations (e.g. scan costs in higher selectivity queries, Haversine distance computation, and point-in-polygon tests), and (v) index lookup is the bottleneck for tree-based structures, which could potentially be reduced by linearizing the indexed partitions.
Human-Centered Autonomy for Autonomous sUAS Target Searching
Authors: Hunter M. Ray, Zakariya Laouar, Zachary Sunberg, Nisar Ahmed
Abstract
Deploying robots that operate in dynamic, uncertain environments, such as Uncrewed Aerial Systems in search \& rescue missions, require nearly continuous human supervision for vehicle guidance and operation. Without approaches that consider high level mission context, operational methods of autonomous flying necessitate cumbersome manual operation or inefficient exhaustive search patterns. To facilitate more effective use of autonomy, we present a human-centered autonomous system that infers geospatial mission context through dynamic features sets, which then guides a probabilistic target search planner. Operators provide a limited set of diverse inputs, including priority definition, spatial semantic observations over ad-hoc geographical areas, and reference waypoints, which are probabilistically fused with geographical database information and condensed into a discretized value map representing an operator's preferences over an operational area. An online, POMDP-based planner, optimized for target searching, is augmented with this value map to generate an operator-constrained vehicle waypoint guidance plan. We validate the system by gathering input from five first responders trained in search \& rescue and compare simulated system performance against current operational methods for autonomous missions. These results display effective task mental model alignment and more efficient guidance plans, resulting in faster rescue times.
Authors: Ivan Grishchenko, Geng Yan, Eduard Gabriel Bazavan, Andrei Zanfir, Nikolai Chinaev, Karthik Raveendran, Matthias Grundmann, Cristian Sminchisescu
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
We present Blendshapes GHUM, an on-device ML pipeline that predicts 52 facial blendshape coefficients at 30+ FPS on modern mobile phones, from a single monocular RGB image and enables facial motion capture applications like virtual avatars. Our main contributions are: i) an annotation-free offline method for obtaining blendshape coefficients from real-world human scans, ii) a lightweight real-time model that predicts blendshape coefficients based on facial landmarks.
SeBaSi system-level Integrated Access and Backhaul simulator for self-backhauling
Authors: Amir Ashtari Gargari, Matteo Pagin, Andrea Ortiz, Nairy Moghadas Gholian, Michele Polese, Michele Zorzi
Subjects: Networking and Internet Architecture (cs.NI)
Abstract
millimeter wave (mmWave) and sub-terahertz (THz) communications have the potential of increasing mobile network throughput drastically. However, the challenging propagation conditions experienced at mmWave and beyond frequencies can potentially limit the range of the wireless link down to a few meters, compared to up to kilometers for sub-6GHz links. Thus, increasing the density of base station deployments is required to achieve sufficient coverage in the Radio Access Network (RAN). To such end, 3rd Generation Partnership Project (3GPP) introduced wireless backhauled base stations with Integrated Access and Backhaul (IAB), a key technology to achieve dense networks while preventing the need for costly fiber deployments. In this paper, we introduce SeBaSi, a system-level simulator for IAB networks, and demonstrate its functionality by simulating IAB deployments in Manhattan, New York City and Padova. Finally, we show how SeBaSi can represent a useful tool for the performance evaluation of self-backhauled cellular networks, thanks to its high level of network abstraction, coupled with its open and customizable design, which allows users to extend it to support novel technologies such as Reconfigurable Intelligent Surfaces (RISs)
Mobile Vision Transformer-based Visual Object Tracking
Authors: Goutam Yelluru Gopal, Maria A. Amer
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
The introduction of robust backbones, such as Vision Transformers, has improved the performance of object tracking algorithms in recent years. However, these state-of-the-art trackers are computationally expensive since they have a large number of model parameters and rely on specialized hardware (e.g., GPU) for faster inference. On the other hand, recent lightweight trackers are fast but are less accurate, especially on large-scale datasets. We propose a lightweight, accurate, and fast tracking algorithm using Mobile Vision Transformers (MobileViT) as the backbone for the first time. We also present a novel approach of fusing the template and search region representations in the MobileViT backbone, thereby generating superior feature encoding for target localization. The experimental results show that our MobileViT-based Tracker, MVT, surpasses the performance of recent lightweight trackers on the large-scale datasets GOT10k and TrackingNet, and with a high inference speed. In addition, our method outperforms the popular DiMP-50 tracker despite having 4.7 times fewer model parameters and running at 2.8 times its speed on a GPU. The tracker code and models are available at https://github.com/goutamyg/MVT
Massive Access of Static and Mobile Users via Reconfigurable Intelligent Surfaces: Protocol Design and Performance Analysis
Authors: Xuelin Cao, Bo Yang, Chongwen Huang, George C. Alexandropoulos, Chau Yuen, Zhu Han, H. Vincent Poor, Lajos Hanzo
Subjects: Networking and Internet Architecture (cs.NI); Signal Processing (eess.SP)
Abstract
The envisioned wireless networks of the future entail the provisioning of massive numbers of connections, heterogeneous data traffic, ultra-high spectral efficiency, and low latency services. This vision is spurring research activities focused on defining a next generation multiple access (NGMA) protocol that can accommodate massive numbers of users in different resource blocks, thereby, achieving higher spectral efficiency and increased connectivity compared to conventional multiple access schemes. In this article, we present a multiple access scheme for NGMA in wireless communication systems assisted by multiple reconfigurable intelligent surfaces (RISs). In this regard, considering the practical scenario of static users operating together with mobile ones, we first study the interplay of the design of NGMA schemes and RIS phase configuration in terms of efficiency and complexity. Based on this, we then propose a multiple access framework for RIS-assisted communication systems, and we also design a medium access control (MAC) protocol incorporating RISs. In addition, we give a detailed performance analysis of the designed RIS-assisted MAC protocol. Our extensive simulation results demonstrate that the proposed MAC design outperforms the benchmarks in terms of system throughput and access fairness, and also reveal a trade-off relationship between the system throughput and fairness.
GVD-Exploration: An Efficient Autonomous Robot Exploration Framework Based on Fast Generalized Voronoi Diagram Extraction
Abstract
Rapidly-exploring Random Trees (RRTs) are a popular technique for autonomous exploration of mobile robots. However, the random sampling used by RRTs can result in inefficient and inaccurate frontiers extraction, which affects the exploration performance. To address the issues of slow path planning and high path cost, we propose a framework that uses a generalized Voronoi diagram (GVD) based multi-choice strategy for robot exploration. Our framework consists of three components: a novel mapping model that uses an end-to-end neural network to construct GVDs of the environments in real time; a GVD-based heuristic scheme that accelerates frontiers extraction and reduces frontiers redundancy; and a multi-choice frontiers assignment scheme that considers different types of frontiers and enables the robot to make rational decisions during the exploration process. We evaluate our method on simulation and real-world experiments and show that it outperforms RRT-based exploration methods in terms of efficiency and robustness.
The first step is the hardest: Pitfalls of Representing and Tokenizing Temporal Data for Large Language Models
Authors: Dimitris Spathis, Fahim Kawsar
Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Abstract
Large Language Models (LLMs) have demonstrated remarkable generalization across diverse tasks, leading individuals to increasingly use them as personal assistants and universal computing engines. Nevertheless, a notable obstacle emerges when feeding numerical/temporal data into these models, such as data sourced from wearables or electronic health records. LLMs employ tokenizers in their input that break down text into smaller units. However, tokenizers are not designed to represent numerical values and might struggle to understand repetitive patterns and context, treating consecutive values as separate tokens and disregarding their temporal relationships. Here, we discuss recent works that employ LLMs for human-centric tasks such as in mobile health sensing and present a case study showing that popular LLMs tokenize temporal data incorrectly. To address that, we highlight potential solutions such as prompt tuning with lightweight embedding layers as well as multimodal adapters, that can help bridge this "modality gap". While the capability of language models to generalize to other modalities with minimal or no finetuning is exciting, this paper underscores the fact that their outputs cannot be meaningful if they stumble over input nuances.
Preliminary Results from a U.S. Demographic Analysis of SMiSh Susceptibility
Authors: Cori Faklaris, Heather Richter Lipford, Sarah Tabassum
Subjects: Human-Computer Interaction (cs.HC); Cryptography and Security (cs.CR); Computers and Society (cs.CY); Social and Information Networks (cs.SI)
Abstract
As adoption of mobile phones has skyrocketed, so have scams involving them. The text method is called SMiShing, (aka SMShing, or smishing) in which a fraudster sends a phishing link via Short Message Service (SMS) text to a phone. However, no data exists on who is most vulnerable to SMiShing. Prior work in phishing (its e-mail cousin) indicates that this is likely to vary by demographic and contextual factors. In our study, we collect this data from N=1007 U.S. adult mobile phone users. Younger people and college students emerge in this sample as the most vulnerable. Participants struggled to correctly identify legitimate messages and were easily misled when they knew they had an account with the faked message entity. Counterintuitively, participants with higher levels of security training and awareness were less correct in rating possible SMiSH. We recommend next steps for researchers, regulators and telecom providers.
Abstract
Can an $\mathbb{R}^n\rightarrow \mathbb{R}^n$ feedforward network learn matrix-vector multiplication? This study introduces two mechanisms - flexible masking to take matrix inputs, and a unique network pruning to respect the mask's dependency structure. Networks can approximate fixed operations such as matrix-vector multiplication $\phi(A,x) \rightarrow Ax$, motivating the mechanisms introduced with applications towards litmus-testing dependencies or interaction order in graph-based models.
Keyword: diffusion
Revisiting Energy Based Models as Policies: Ranking Noise Contrastive Estimation and Interpolating Energy Models
Authors: Sumeet Singh, Stephen Tu, Vikas Sindhwani
Abstract
A crucial design decision for any robot learning pipeline is the choice of policy representation: what type of model should be used to generate the next set of robot actions? Owing to the inherent multi-modal nature of many robotic tasks, combined with the recent successes in generative modeling, researchers have turned to state-of-the-art probabilistic models such as diffusion models for policy representation. In this work, we revisit the choice of energy-based models (EBM) as a policy class. We show that the prevailing folklore -- that energy models in high dimensional continuous spaces are impractical to train -- is false. We develop a practical training objective and algorithm for energy models which combines several key ingredients: (i) ranking noise contrastive estimation (R-NCE), (ii) learnable negative samplers, and (iii) non-adversarial joint training. We prove that our proposed objective function is asymptotically consistent and quantify its limiting variance. On the other hand, we show that the Implicit Behavior Cloning (IBC) objective is actually biased even at the population level, providing a mathematical explanation for the poor performance of IBC trained energy policies in several independent follow-up works. We further extend our algorithm to learn a continuous stochastic process that bridges noise and data, modeling this process with a family of EBMs indexed by scale variable. In doing so, we demonstrate that the core idea behind recent progress in generative modeling is actually compatible with EBMs. Altogether, our proposed training algorithms enable us to train energy-based models as policies which compete with -- and even outperform -- diffusion models and other state-of-the-art approaches in several challenging multi-modal benchmarks: obstacle avoidance path planning and contact-rich block pushing.
Catch You Everything Everywhere: Guarding Textual Inversion via Concept Watermarking
Abstract
AIGC (AI-Generated Content) has achieved tremendous success in many applications such as text-to-image tasks, where the model can generate high-quality images with diverse prompts, namely, different descriptions in natural languages. More surprisingly, the emerging personalization techniques even succeed in describing unseen concepts with only a few personal images as references, and there have been some commercial platforms for sharing the valuable personalized concept. However, such an advanced technique also introduces a severe threat, where malicious users can misuse the target concept to generate highly-realistic illegal images. Therefore, it becomes necessary for the platform to trace malicious users and hold them accountable. In this paper, we focus on guarding the most popular lightweight personalization model, ie, Textual Inversion (TI). To achieve it, we propose the novel concept watermarking, where watermark information is embedded into the target concept and then extracted from generated images based on the watermarked concept. Specifically, we jointly train a watermark encoder and a watermark decoder with the sampler in the loop. It shows great resilience to different diffusion sampling processes possibly chosen by malicious users, meanwhile preserving utility for normal use. In practice, the concept owner can upload his concept with different watermarks (ie, serial numbers) to the platform, and the platform allocates different users with different serial numbers for subsequent tracing and forensics.
Beyond Generation: Harnessing Text to Image Models for Object Detection and Segmentation
Abstract
We propose a new paradigm to automatically generate training data with accurate labels at scale using the text-to-image synthesis frameworks (e.g., DALL-E, Stable Diffusion, etc.). The proposed approach1 decouples training data generation into foreground object generation, and contextually coherent background generation. To generate foreground objects, we employ a straightforward textual template, incorporating the object class name as input prompts. This is fed into a text-to-image synthesis framework, producing various foreground images set against isolated backgrounds. A foreground-background segmentation algorithm is then used to generate foreground object masks. To generate context images, we begin by creating language descriptions of the context. This is achieved by applying an image captioning method to a small set of images representing the desired context. These textual descriptions are then transformed into a diverse array of context images via a text-to-image synthesis framework. Subsequently, we composite these with the foreground object masks produced in the initial step, utilizing a cut-and-paste method, to formulate the training data. We demonstrate the advantages of our approach on five object detection and segmentation datasets, including Pascal VOC and COCO. We found that detectors trained solely on synthetic data produced by our method achieve performance comparable to those trained on real data (Fig. 1). Moreover, a combination of real and synthetic data yields even much better results. Further analysis indicates that the synthetic data distribution complements the real data distribution effectively. Additionally, we emphasize the compositional nature of our data generation approach in out-of-distribution and zero-shot data generation scenarios. We open-source our code at https://github.com/gyhandy/Text2Image-for-Detection
Prompting4Debugging: Red-Teaming Text-to-Image Diffusion Models by Finding Problematic Prompts
Abstract
Text-to-image diffusion models, e.g. Stable Diffusion (SD), lately have shown remarkable ability in high-quality content generation, and become one of the representatives for the recent wave of transformative AI. Nevertheless, such advance comes with an intensifying concern about the misuse of this generative technology, especially for producing copyrighted or NSFW (i.e. not safe for work) images. Although efforts have been made to filter inappropriate images/prompts or remove undesirable concepts/styles via model fine-tuning, the reliability of these safety mechanisms against diversified problematic prompts remains largely unexplored. In this work, we propose Prompting4Debugging (P4D) as a debugging and red-teaming tool that automatically finds problematic prompts for diffusion models to test the reliability of a deployed safety mechanism. We demonstrate the efficacy of our P4D tool in uncovering new vulnerabilities of SD models with safety mechanisms. Particularly, our result shows that around half of prompts in existing safe prompting benchmarks which were originally considered "safe" can actually be manipulated to bypass many deployed safety mechanisms, including concept removal, negative prompt, and safety guidance. Our findings suggest that, without comprehensive testing, the evaluations on limited safe prompting benchmarks can lead to a false sense of safety for text-to-image models.
Elucidating the solution space of extended reverse-time SDE for diffusion models
Abstract
Diffusion models (DMs) demonstrate potent image generation capabilities in various generative modeling tasks. Nevertheless, their primary limitation lies in slow sampling speed, requiring hundreds or thousands of sequential function evaluations through large neural networks to generate high-quality images. Sampling from DMs can be seen as solving corresponding stochastic differential equations (SDEs) or ordinary differential equations (ODEs). In this work, we formulate the sampling process as an extended reverse-time SDE (ER SDE), unifying prior explorations into ODEs and SDEs. Leveraging the semi-linear structure of ER SDE solutions, we offer exact solutions and arbitrarily high-order approximate solutions for VP SDE and VE SDE, respectively. Based on the solution space of the ER SDE, we yield mathematical insights elucidating the superior performance of ODE solvers over SDE solvers in terms of fast sampling. Additionally, we unveil that VP SDE solvers stand on par with their VE SDE counterparts. Finally, we devise fast and training-free samplers, ER-SDE Solvers, elevating the efficiency of stochastic samplers to unprecedented levels. Experimental results demonstrate achieving 3.45 FID in 20 function evaluations and 2.24 FID in 50 function evaluations on the ImageNet 64$\times$64 dataset.
InstaFlow: One Step is Enough for High-Quality Diffusion-Based Text-to-Image Generation
Abstract
Diffusion models have revolutionized text-to-image generation with its exceptional quality and creativity. However, its multi-step sampling process is known to be slow, often requiring tens of inference steps to obtain satisfactory results. Previous attempts to improve its sampling speed and reduce computational costs through distillation have been unsuccessful in achieving a functional one-step model. In this paper, we explore a recent method called Rectified Flow, which, thus far, has only been applied to small datasets. The core of Rectified Flow lies in its \emph{reflow} procedure, which straightens the trajectories of probability flows, refines the coupling between noises and images, and facilitates the distillation process with student models. We propose a novel text-conditioned pipeline to turn Stable Diffusion (SD) into an ultra-fast one-step model, in which we find reflow plays a critical role in improving the assignment between noise and images. Leveraging our new pipeline, we create, to the best of our knowledge, the first one-step diffusion-based text-to-image generator with SD-level image quality, achieving an FID (Frechet Inception Distance) of $23.3$ on MS COCO 2017-5k, surpassing the previous state-of-the-art technique, progressive distillation, by a significant margin ($37.2$ $\rightarrow$ $23.3$ in FID). By utilizing an expanded network with 1.7B parameters, we further improve the FID to $22.4$. We call our one-step models \emph{InstaFlow}. On MS COCO 2014-30k, InstaFlow yields an FID of $13.1$ in just $0.09$ second, the best in $\leq 0.1$ second regime, outperforming the recent StyleGAN-T ($13.9$ in $0.1$ second). Notably, the training of InstaFlow only costs 199 A100 GPU days. Project page:~\url{https://github.com/gnobitab/InstaFlow}.
A Fast Algorithm for Moderating Critical Nodes via Edge Removal
Authors: Changan Liu, Xiaotian Zhou, Ahad N. Zehmakan, Zhongzhi Zhang
Subjects: Social and Information Networks (cs.SI); Artificial Intelligence (cs.AI)
Abstract
Critical nodes in networks are extremely vulnerable to malicious attacks to trigger negative cascading events such as the spread of misinformation and diseases. Therefore, effective moderation of critical nodes is very vital for mitigating the potential damages caused by such malicious diffusions. The current moderation methods are computationally expensive. Furthermore, they disregard the fundamental metric of information centrality, which measures the dissemination power of nodes. We investigate the problem of removing $k$ edges from a network to minimize the information centrality of a target node $\lea$ while preserving the network's connectivity. We prove that this problem is computationally challenging: it is NP-complete and its objective function is not supermodular. However, we propose three approximation greedy algorithms using novel techniques such as random walk-based Schur complement approximation and fast sum estimation. One of our algorithms runs in nearly linear time in the number of edges. To complement our theoretical analysis, we conduct a comprehensive set of experiments on synthetic and real networks with over one million nodes. Across various settings, the experimental results illustrate the effectiveness and efficiency of our proposed algorithms.
Keyword: adaptive
SHAPE: A Sample-adaptive Hierarchical Prediction Network for Medication Recommendation
Authors: Sicen Liu, Xiaolong Wang, JIngcheng Du, Yongshuai Hou, Xianbing Zhao, Hui Xu, Hui Wang, Yang Xiang, Buzhou Tang
Abstract
Effectively medication recommendation with complex multimorbidity conditions is a critical task in healthcare. Most existing works predicted medications based on longitudinal records, which assumed the information transmitted patterns of learning longitudinal sequence data are stable and intra-visit medical events are serialized. However, the following conditions may have been ignored: 1) A more compact encoder for intra-relationship in the intra-visit medical event is urgent; 2) Strategies for learning accurate representations of the variable longitudinal sequences of patients are different. In this paper, we proposed a novel Sample-adaptive Hierarchical medicAtion Prediction nEtwork, termed SHAPE, to tackle the above challenges in the medication recommendation task. Specifically, we design a compact intra-visit set encoder to encode the relationship in the medical event for obtaining visit-level representation and then develop an inter-visit longitudinal encoder to learn the patient-level longitudinal representation efficiently. To endow the model with the capability of modeling the variable visit length, we introduce a soft curriculum learning method to assign the difficulty of each sample automatically by the visit length. Extensive experiments on a benchmark dataset verify the superiority of our model compared with several state-of-the-art baselines.
Online ML Self-adaptation in Face of Traps
Authors: Michal Töpfer, František Plášil, Tomáš Bureš, Petr Hnětynka, Martin Kruliš, Danny Weyns
Subjects: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
Abstract
Online machine learning (ML) is often used in self-adaptive systems to strengthen the adaptation mechanism and improve the system utility. Despite such benefits, applying online ML for self-adaptation can be challenging, and not many papers report its limitations. Recently, we experimented with applying online ML for self-adaptation of a smart farming scenario and we had faced several unexpected difficulties -- traps -- that, to our knowledge, are not discussed enough in the community. In this paper, we report our experience with these traps. Specifically, we discuss several traps that relate to the specification and online training of the ML-based estimators, their impact on self-adaptation, and the approach used to evaluate the estimators. Our overview of these traps provides a list of lessons learned, which can serve as guidance for other researchers and practitioners when applying online ML for self-adaptation.
Concurrent Composition for Interactive Differential Privacy with Adaptive Privacy-Loss Parameters
Authors: Samuel Haney, Michael Shoemate, Grace Tian, Salil Vadhan, Andrew Vyrros, Vicki Xu, Wanrong Zhang
Subjects: Cryptography and Security (cs.CR); Data Structures and Algorithms (cs.DS); Information Theory (cs.IT)
Abstract
In this paper, we study the concurrent composition of interactive mechanisms with adaptively chosen privacy-loss parameters. In this setting, the adversary can interleave queries to existing interactive mechanisms, as well as create new ones. We prove that every valid privacy filter and odometer for noninteractive mechanisms extends to the concurrent composition of interactive mechanisms if privacy loss is measured using $(\epsilon, \delta)$-DP, $f$-DP, or R\'enyi DP of fixed order. Our results offer strong theoretical foundations for enabling full adaptivity in composing differentially private interactive mechanisms, showing that concurrency does not affect the privacy guarantees. We also provide an implementation for users to deploy in practice.
Life-inspired Interoceptive Artificial Intelligence for Autonomous and Adaptive Agents
Authors: Sungwoo Lee, Younghyun Oh, Hyunhoe An, Hyebhin Yoon, Karl J. Friston, Seok Jun Hong, Choong-Wan Woo
Subjects: Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE)
Abstract
Building autonomous --- i.e., choosing goals based on one's needs -- and adaptive -- i.e., surviving in ever-changing environments -- agents has been a holy grail of artificial intelligence (AI). A living organism is a prime example of such an agent, offering important lessons about adaptive autonomy. Here, we focus on interoception, a process of monitoring one's internal environment to keep it within certain bounds, which underwrites the survival of an organism. To develop AI with interoception, we need to factorize the state variables representing internal environments from external environments and adopt life-inspired mathematical properties of internal environment states. This paper offers a new perspective on how interoception can help build autonomous and adaptive agents by integrating the legacy of cybernetics with recent advances in theories of life, reinforcement learning, and neuroscience.
Learning from History: Task-agnostic Model Contrastive Learning for Image Restoration
Authors: Gang Wu, Junjun Jiang, Kui Jiang, Xianming Liu
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Contrastive learning has emerged as a prevailing paradigm for high-level vision tasks, which, by introducing properly negative samples, has also been exploited for low-level vision tasks to achieve a compact optimization space to account for their ill-posed nature. However, existing methods rely on manually predefined, task-oriented negatives, which often exhibit pronounced task-specific biases. In this paper, we propose a innovative approach for the adaptive generation of negative samples directly from the target model itself, called learning from history. We introduce the Self-Prior guided Negative loss for image restoration (SPNIR) to enable this approach. Our approach is task-agnostic and generic, making it compatible with any existing image restoration method or task. We demonstrate the effectiveness of our approach by retraining existing models with SPNIR. The results show significant improvements in image restoration across various tasks and architectures. For example, models retrained with SPNIR outperform the original FFANet and DehazeFormer by 3.41 dB and 0.57 dB on the RESIDE indoor dataset for image dehazing. Similarly, they achieve notable improvements of 0.47 dB on SPA-Data over IDT for image deraining and 0.12 dB on Manga109 for a 4x scale super-resolution over lightweight SwinIR, respectively. Code and retrained models are available at https://github.com/Aitical/Task-agnostic_Model_Contrastive_Learning_Image_Restoration.
Accelerating Edge AI with Morpher: An Integrated Design, Compilation and Simulation Framework for CGRAs
Abstract
Coarse-Grained Reconfigurable Arrays (CGRAs) hold great promise as power-efficient edge accelerator, offering versatility beyond AI applications. Morpher, an open-source, architecture-adaptive CGRA design framework, is specifically designed to explore the vast design space of CGRAs. The comprehensive ecosystem of Morpher includes a tailored compiler, simulator, accelerator synthesis, and validation framework. This study provides an overview of Morpher, highlighting its capabilities in automatically compiling AI application kernels onto user-defined CGRA architectures and verifying their functionality. Through the Morpher framework, the versatility of CGRAs is harnessed to facilitate efficient compilation and verification of edge AI applications, covering important kernels representative of a wide range of embedded AI workloads. Morpher is available online at https://github.com/ecolab-nus/morpher-v2.
Adaptive Bregman-Kaczmarz: An Approach to Solve Linear Inverse Problems with Independent Noise Exactly
Authors: Lionel Tondji, Idriss Tondji, Dirk A. Lorenz
Subjects: Numerical Analysis (math.NA); Optimization and Control (math.OC)
Abstract
We consider the block Bregman-Kaczmarz method for finite dimensional linear inverse problems. The block Bregman-Kaczmarz method uses blocks of the linear system and performs iterative steps with these blocks only. We assume a noise model that we call independent noise, i.e. each time the method performs a step for some block, one obtains a noisy sample of the respective part of the right-hand side which is contaminated with new noise that is independent of all previous steps of the method. One can view these noise models as making a fresh noisy measurement of the respective block each time it is used. In this framework, we are able to show that a well-chosen adaptive stepsize of the block Bergman-Kaczmarz method is able to converge to the exact solution of the linear inverse problem. The plain form of this adaptive stepsize relies on unknown quantities (like the Bregman distance to the solution), but we show a way how these quantities can be estimated purely from given data. We illustrate the finding in numerical experiments and confirm that these heuristic estimates lead to effective stepsizes.
SAMPLING: Scene-adaptive Hierarchical Multiplane Images Representation for Novel View Synthesis from a Single Image
Abstract
Recent novel view synthesis methods obtain promising results for relatively small scenes, e.g., indoor environments and scenes with a few objects, but tend to fail for unbounded outdoor scenes with a single image as input. In this paper, we introduce SAMPLING, a Scene-adaptive Hierarchical Multiplane Images Representation for Novel View Synthesis from a Single Image based on improved multiplane images (MPI). Observing that depth distribution varies significantly for unbounded outdoor scenes, we employ an adaptive-bins strategy for MPI to arrange planes in accordance with each scene image. To represent intricate geometry and multi-scale details, we further introduce a hierarchical refinement branch, which results in high-quality synthesized novel views. Our method demonstrates considerable performance gains in synthesizing large-scale unbounded outdoor scenes using a single image on the KITTI dataset and generalizes well to the unseen Tanks and Temples dataset. The code and models will be made public.
A New Re-redistribution Scheme for Weighted State Redistribution with Adaptive Mesh Refinement
Authors: Isabel Barrio Sanchez, Ann S. Almgren, John B. Bell, Marc T. Henry de Frahan, Weiqun Zhang
Abstract
State redistribution (SRD) is a recently developed technique for stabilizing cut cells that result from finite-volume embedded boundary methods. SRD has been successfully applied to a variety of compressible and incompressible flow problems. When used in conjunction with adaptive mesh refinement (AMR), additional steps are needed to preserve the accuracy and conservation properties of the solution if the embedded boundary is not restricted to a single level of the mesh hierarchy. In this work, we extend the weighted state redistribution algorithm to cases where cut cells live at or near a coarse-fine interface within the domain. The resulting algorithm maintains conservation and is demonstrated on several two- and three-dimensional example problems.
Keyword: efficient
SHAPE: A Sample-adaptive Hierarchical Prediction Network for Medication Recommendation
Temporal Patience: Efficient Adaptive Deep Learning for Embedded Radar Data Processing
Blendshapes GHUM: Real-time Monocular Facial Blendshape Prediction
Grey-box Bayesian Optimization for Sensor Placement in Assisted Living Environments
PACE: Prompting and Augmentation for Calibrated Confidence Estimation with GPT-4 in Cloud Incident Root Cause Analysis
The Safety Filter: A Unified View of Safety-Critical Control in Autonomous Systems
Fairness- and uncertainty-aware data generation for data-driven design
ChemSpaceAL: An Efficient Active Learning Methodology Applied to Protein-Specific Molecular Generation
Uncovering mesa-optimization algorithms in Transformers
Force-directed graph embedding with hops distance
SkillScanner: Detecting Policy-Violating Voice Applications Through Static Analysis at the Development Phase
Systemization of Knowledge (SoK)- Cross Impact of Transfer Learning in Cybersecurity: Offensive, Defensive and Threat Intelligence Perspectives
Fast Constraint Screening for Multi-Interval Unit Commitment
Geometry Enhanced Optimal Control Technique for Acrobatic Flip Motion of Quadcopter
Reset Controller Synthesis by Reach-avoid Analysis for Delay Hybrid Systems
Combining deep learning and street view imagery to map smallholder crop types
Random Segmentation: New Traffic Obfuscation against Packet-Size-Based Side-Channel Attacks
Language Models as Black-Box Optimizers for Vision-Language Models
Trust-Region Neural Moving Horizon Estimation for Robots
A correction function-based kernel-free boundary integral method for elliptic PDEs with implicitly defined interfaces
Gait Design of a Novel Arboreal Concertina Locomotion for Snake-like Robots
GVD-Exploration: An Efficient Autonomous Robot Exploration Framework Based on Fast Generalized Voronoi Diagram Extraction
Update Monte Carlo tree search (UMCTS) algorithm for heuristic global search of sizing optimization problems for truss structures
Real-Time Semantic Segmentation: A Brief Survey & Comparative Study in Remote Sensing
A Perceptron-based Fine Approximation Technique for Linear Separation
OmniSketch: Efficient Multi-Dimensional High-Velocity Stream Analytics with Arbitrary Predicates
Information Flow in Graph Neural Networks: A Clinical Triage Use Case
Measuring Catastrophic Forgetting in Cross-Lingual Transfer Paradigms: Exploring Tuning Strategies
HOC-Search: Efficient CAD Model and Pose Retrieval from RGB-D Scans
Inspection planning under execution uncertainty
Dynamic Visual Prompt Tuning for Parameter Efficient Transfer Learning
Accelerating Edge AI with Morpher: An Integrated Design, Compilation and Simulation Framework for CGRAs
Active Label Refinement for Semantic Segmentation of Satellite Images
Integration of Quantum Accelerators with High Performance Computing $\unicode{x2013}$ A Review of Quantum Programming Tools
AKEM: Aligning Knowledge Base to Queries with Ensemble Model for Entity Recognition and Linking
Efficient Memory Management for Large Language Model Serving with PagedAttention
360$^\circ$ from a Single Camera: A Few-Shot Approach for LiDAR Segmentation
High Order Numerical Methods To Approximate The Singular Value Decomposition
SGFeat: Salient Geometric Feature for Point Cloud Registration
Optimal Quota for a Multi-species Fishing Models
Speciality vs Generality: An Empirical Study on Catastrophic Forgetting in Fine-tuning Foundation Models
OTAS: Unsupervised Boundary Detection for Object-Centric Temporal Action Segmentation
Transferability analysis of data-driven additive manufacturing knowledge: a case study between powder bed fusion and directed energy deposition
Using conservative voltage reduction and dynamic thermal rating for congestion management of power network
Distributed Precoding for Satellite-Terrestrial Integrated Networks Without Sharing CSIT: A Rate-Splitting Approach
A Simple Multiple-Access Design for Reconfigurable Intelligent Surface-Aided Systems
Enhancing In-Memory Spatial Indexing with Learned Search
Human-Centered Autonomy for Autonomous sUAS Target Searching
On Computationally Efficient Learning of Exponential Family Distributions
C4CAM: A Compiler for CAM-based In-memory Accelerators
Keyword: faster
Mobile Vision Transformer-based Visual Object Tracking
Analysis of Randomized Householder-Cholesky QR Factorization with Multisketching
Update Monte Carlo tree search (UMCTS) algorithm for heuristic global search of sizing optimization problems for truss structures
Efficient Memory Management for Large Language Model Serving with PagedAttention
The Time Complexity of Fully Sparse Matrix Multiplication
Enhancing In-Memory Spatial Indexing with Learned Search
Human-Centered Autonomy for Autonomous sUAS Target Searching
Keyword: mobile
Blendshapes GHUM: Real-time Monocular Facial Blendshape Prediction
SeBaSi system-level Integrated Access and Backhaul simulator for self-backhauling
Mobile Vision Transformer-based Visual Object Tracking
Massive Access of Static and Mobile Users via Reconfigurable Intelligent Surfaces: Protocol Design and Performance Analysis
GVD-Exploration: An Efficient Autonomous Robot Exploration Framework Based on Fast Generalized Voronoi Diagram Extraction
The first step is the hardest: Pitfalls of Representing and Tokenizing Temporal Data for Large Language Models
Preliminary Results from a U.S. Demographic Analysis of SMiSh Susceptibility
Keyword: pruning
Ensemble Mask Networks
Keyword: diffusion
Revisiting Energy Based Models as Policies: Ranking Noise Contrastive Estimation and Interpolating Energy Models
Catch You Everything Everywhere: Guarding Textual Inversion via Concept Watermarking
Beyond Generation: Harnessing Text to Image Models for Object Detection and Segmentation
Prompting4Debugging: Red-Teaming Text-to-Image Diffusion Models by Finding Problematic Prompts
Elucidating the solution space of extended reverse-time SDE for diffusion models
InstaFlow: One Step is Enough for High-Quality Diffusion-Based Text-to-Image Generation
A Fast Algorithm for Moderating Critical Nodes via Edge Removal
Keyword: adaptive
SHAPE: A Sample-adaptive Hierarchical Prediction Network for Medication Recommendation
Online ML Self-adaptation in Face of Traps
Concurrent Composition for Interactive Differential Privacy with Adaptive Privacy-Loss Parameters
Life-inspired Interoceptive Artificial Intelligence for Autonomous and Adaptive Agents
Learning from History: Task-agnostic Model Contrastive Learning for Image Restoration
learning from history
. We introduce the Self-Prior guided Negative loss for image restoration (SPNIR) to enable this approach. Our approach is task-agnostic and generic, making it compatible with any existing image restoration method or task. We demonstrate the effectiveness of our approach by retraining existing models with SPNIR. The results show significant improvements in image restoration across various tasks and architectures. For example, models retrained with SPNIR outperform the original FFANet and DehazeFormer by 3.41 dB and 0.57 dB on the RESIDE indoor dataset for image dehazing. Similarly, they achieve notable improvements of 0.47 dB on SPA-Data over IDT for image deraining and 0.12 dB on Manga109 for a 4x scale super-resolution over lightweight SwinIR, respectively. Code and retrained models are available at https://github.com/Aitical/Task-agnostic_Model_Contrastive_Learning_Image_Restoration.Accelerating Edge AI with Morpher: An Integrated Design, Compilation and Simulation Framework for CGRAs
Adaptive Bregman-Kaczmarz: An Approach to Solve Linear Inverse Problems with Independent Noise Exactly
SAMPLING: Scene-adaptive Hierarchical Multiplane Images Representation for Novel View Synthesis from a Single Image
A New Re-redistribution Scheme for Weighted State Redistribution with Adaptive Mesh Refinement
Keyword: quantization
There is no result