Abstract
Neural Radiance Field (NeRF) models have gained significant attention in the computer vision community in the recent past with state-of-the-art visual quality and produced impressive demonstrations. Since then, technopreneurs have sought to leverage NeRF models into a profitable business. Therefore, NeRF models make it worth the risk of plagiarizers illegally copying, re-distributing, or misusing those models. This paper proposes a comprehensive intellectual property (IP) protection framework for the NeRF model in both black-box and white-box settings, namely IPR-NeRF. In the black-box setting, a diffusion-based solution is introduced to embed and extract the watermark via a two-stage optimization process. In the white-box setting, a designated digital signature is embedded into the weights of the NeRF model by adopting the sign loss objective. Our extensive experiments demonstrate that not only does our approach maintain the fidelity (\ie, the rendering quality) of IPR-NeRF models, but it is also robust against both ambiguity and removal attacks compared to prior arts.
Aligning Large Language Models with Counterfactual DPO
Authors: Authors: Bradley Butcher
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Abstract
Advancements in large language models (LLMs) have demonstrated remarkable capabilities across a diverse range of applications. These models excel in generating text completions that are contextually coherent and cover an extensive array of subjects. However, the vast datasets required for their training make aligning response styles during the pretraining and instruction tuning phases challenging. Consequently, an additional alignment phase is typically employed, wherein the model is further trained with human preference data to better align its outputs with human expectations. While this process doesn't introduce new capabilities per se, it does accentuate generation styles innate to the model. This paper explores the utilization of counterfactual prompting within the framework of Direct Preference Optimization (DPO) to align the model's style without relying on human intervention. We demonstrate that this method effectively instils desirable behaviour, mitigates undesirable ones, and encourages the model to disregard inappropriate instructions. Our findings suggest that counterfactual prompting with DPO presents a low-resource way to fine-tune LLMs to meet the demands for responsible and ethically aligned AI systems.
Bilevel Optimization under Unbounded Smoothness: A New Algorithm and Convergence Analysis
Authors: Authors: Jie Hao, Xiaochuan Gong, Mingrui Liu
Subjects: Machine Learning (cs.LG); Optimization and Control (math.OC)
Abstract
Bilevel optimization is an important formulation for many machine learning problems. Current bilevel optimization algorithms assume that the gradient of the upper-level function is Lipschitz. However, recent studies reveal that certain neural networks such as recurrent neural networks (RNNs) and long-short-term memory networks (LSTMs) exhibit potential unbounded smoothness, rendering conventional bilevel optimization algorithms unsuitable. In this paper, we design a new bilevel optimization algorithm, namely BO-REP, to address this challenge. This algorithm updates the upper-level variable using normalized momentum and incorporates two novel techniques for updating the lower-level variable: \textit{initialization refinement} and \textit{periodic updates}. Specifically, once the upper-level variable is initialized, a subroutine is invoked to obtain a refined estimate of the corresponding optimal lower-level variable, and the lower-level variable is updated only after every specific period instead of each iteration. When the upper-level problem is nonconvex and unbounded smooth, and the lower-level problem is strongly convex, we prove that our algorithm requires $\widetilde{\mathcal{O}}(1/\epsilon^4)$ iterations to find an $\epsilon$-stationary point in the stochastic setting, where each iteration involves calling a stochastic gradient or Hessian-vector product oracle. Notably, this result matches the state-of-the-art complexity results under the bounded smoothness setting and without mean-squared smoothness of the stochastic gradient, up to logarithmic factors. Our proof relies on novel technical lemmas for the periodically updated lower-level variable, which are of independent interest. Our experiments on hyper-representation learning, hyperparameter optimization, and data hyper-cleaning for text classification tasks demonstrate the effectiveness of our proposed algorithm.
Transient dynamics under structured perturbations: bridging unstructured and structured pseudospectra
Authors: Authors: Christian Lubich, Nicola Guglielmi
Abstract
The structured $\varepsilon$-stability radius is introduced as a quantity to assess the robustness of transient bounds of solutions to linear differential equations under structured perturbations of the matrix. This applies to general linear structures such as complex or real matrices with a given sparsity pattern or with restricted range and corange, or special classes such as Toeplitz matrices. The notion conceptually combines unstructured and structured pseudospectra in a joint pseudospectrum, allowing for the use of resolvent bounds as with unstructured pseudospectra and for structured perturbations as with structured pseudospectra. We propose and study an algorithm for computing the structured $\varepsilon$-stability radius. This algorithm solves eigenvalue optimization problems via suitably discretized rank-1 matrix differential equations that originate from a gradient system. The proposed algorithm has essentially the same computational cost as the known rank-1 algorithms for computing unstructured and structured stability radii. Numerical experiments illustrate the behavior of the algorithm.
SMOOTHIE: A Theory of Hyper-parameter Optimization for Software Analytics
Abstract
Hyper-parameter optimization is the black art of tuning a learner's control parameters. In software analytics, a repeated result is that such tuning can result in dramatic performance improvements. Despite this, hyper-parameter optimization is often applied rarely or poorly in software analytics--perhaps due to the CPU cost of exploring all those parameter options can be prohibitive. We theorize that learners generalize better when the loss landscape is smooth''. This theory is useful since the influence onsmoothness'' of different hyper-parameter choices can be tested very quickly (e.g. for a deep learner, after just one epoch). To test this theory, this paper implements and tests SMOOTHIE, a novel hyper-parameter optimizer that guides its optimizations via considerations of smothness''. The experiments of this paper test SMOOTHIE on numerous SE tasks including (a) GitHub issue lifetime prediction; (b) detecting false alarms in static code warnings; (c) defect prediction, and (d) a set of standard ML datasets. In all these experiments, SMOOTHIE out-performed state-of-the-art optimizers. Better yet, SMOOTHIE ran 300% faster than the prior state-of-the art. We hence conclude that this theory (that hyper-parameter optimization is best viewed as asmoothing'' function for the decision landscape), is both theoretically interesting and practically very useful. To support open science and other researchers working in this area, all our scripts and datasets are available on-line at https://github.com/yrahul3910/smoothness-hpo/.
Abstract
In this paper we propose a new non-linear classifier based on a combination of locally linear classifiers. A well known optimization formulation is given as we cast the problem in a $\ell_1$ Multiple Kernel Learning (MKL) problem using many locally linear kernels. Since the number of such kernels is huge, we provide a scalable generic MKL training algorithm handling streaming kernels. With respect to the inference time, the resulting classifier fits the gap between high accuracy but slow non-linear classifiers (such as classical MKL) and fast but low accuracy linear classifiers.
Convex and Bilevel Optimization for Neuro-Symbolic Inference and Learning
Authors: Authors: Charles Dickens, Changyu Gao, Connor Pryor, Stephen Wright, Lise Getoor
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Optimization and Control (math.OC)
Abstract
We address a key challenge for neuro-symbolic (NeSy) systems by leveraging convex and bilevel optimization techniques to develop a general gradient-based framework for end-to-end neural and symbolic parameter learning. The applicability of our framework is demonstrated with NeuPSL, a state-of-the-art NeSy architecture. To achieve this, we propose a smooth primal and dual formulation of NeuPSL inference and show learning gradients are functions of the optimal dual variables. Additionally, we develop a dual block coordinate descent algorithm for the new formulation that naturally exploits warm-starts. This leads to over 100x learning runtime improvements over the current best NeuPSL inference method. Finally, we provide extensive empirical evaluations across $8$ datasets covering a range of tasks and demonstrate our learning framework achieves up to a 16% point prediction performance improvement over alternative learning methods.
QoS-Aware 3D Coverage Deployment of UAVs for Internet of Vehicles in Intelligent Transportation
Authors: Authors: engfei Du, Tingyue Xiao, Haotong Cao, Daosen Zhai
Abstract
It is a challenging problem to characterize the air-to-ground (A2G) channel and identify the best deployment location for 3D UAVs with the QoS awareness. To address this problem, we propose a QoS-aware UAV 3D coverage deployment algorithm, which simulates the three-dimensional urban road scenario, considers the UAV communication resource capacity and vehicle communication QoS requirements comprehensively, and then obtains the optimal UAV deployment position by improving the genetic algorithm. Specifically, the K-means clustering algorithm is used to cluster the vehicles, and the center locations of these clusters serve as the initial UAV positions to generate the initial population. Subsequently, we employ the K-means initialized grey wolf optimization (KIGWO) algorithm to achieve the UAV location with an optimal fitness value by performing an optimal search within the grey wolf population. To enhance the algorithm's diversity and global search capability, we randomly substitute this optimal location with one of the individual locations from the initial population. The fitness value is determined by the total number of vehicles covered by UAVs in the system, while the allocation scheme's feasibility is evaluated based on the corresponding QoS requirements. Competitive selection operations are conducted to retain individuals with higher fitness values, while crossover and mutation operations are employed to maintain the diversity of solutions. Finally, the individual with the highest fitness, which represents the UAV deployment position that covers the maximum number of vehicles in the entire system, is selected as the optimal solution. Extensive experimental results demonstrate that the proposed algorithm can effectively enhance the reliability and vehicle communication QoS.
Joint Beam Direction Control and Radio Resource Allocation in Dynamic Multi-beam LEO Satellite Networks
Abstract
Multi-beam low earth orbit (LEO) satellites are emerging as key components in beyond 5G and 6G to provide global coverage and high data rate. To fully unleash the potential of LEO satellite communication, resource management plays a key role. However, the uneven distribution of users, the coupling of multi-dimensional resources, complex inter-beam interference, and time-varying network topologies all impose significant challenges on effective communication resource management. In this paper, we study the joint optimization of beam direction and the allocation of spectrum, time, and power resource in a dynamic multi-beam LEO satellite network. The objective is to improve long-term user sum data rate while taking user fairness into account. Since the concerned resource management problem is mixed-integer non-convex programming, the problem is decomposed into three subproblems, namely beam direction control and time slot allocation, user subchannel assignment, and beam power allocation. Then, these subproblems are solved iteratively by leveraging matching with externalities and successive convex approximation, and the proposed algorithms are analyzed in terms of stability, convergence, and complexity. Extensive simulations are conducted, and the results demonstrate that our proposal can improve the number of served users by up to two times and the sum user data rate by up to 68%, compared to baseline schemes.
Measuring the Discrepancy between 3D Geometric Models using Directional Distance Fields
Abstract
Qualifying the discrepancy between 3D geometric models, which could be represented with either point clouds or triangle meshes, is a pivotal issue with board applications. Existing methods mainly focus on directly establishing the correspondence between two models and then aggregating point-wise distance between corresponding points, resulting in them being either inefficient or ineffective. In this paper, we propose DirDist, an efficient, effective, robust, and differentiable distance metric for 3D geometry data. Specifically, we construct DirDist based on the proposed implicit representation of 3D models, namely directional distance field (DDF), which defines the directional distances of 3D points to a model to capture its local surface geometry. We then transfer the discrepancy between two 3D geometric models as the discrepancy between their DDFs defined on an identical domain, naturally establishing model correspondence. To demonstrate the advantage of our DirDist, we explore various distance metric-driven 3D geometric modeling tasks, including template surface fitting, rigid registration, non-rigid registration, scene flow estimation and human pose optimization. Extensive experiments show that our DirDist achieves significantly higher accuracy under all tasks. As a generic distance metric, DirDist has the potential to advance the field of 3D geometric modeling. The source code is available at \url{https://github.com/rsy6318/DirDist}.
Applications of Machine Learning to Optimizing Polyolefin Manufacturing
Authors: Authors: Niket Sharma, Y.A. Liu
Subjects: Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE)
Abstract
This chapter is a preprint from our book by , focusing on leveraging machine learning (ML) in chemical and polyolefin manufacturing optimization. It's crafted for both novices and seasoned professionals keen on the latest ML applications in chemical processes. We trace the evolution of AI and ML in chemical industries, delineate core ML components, and provide resources for ML beginners. A detailed discussion on various ML methods is presented, covering regression, classification, and unsupervised learning techniques, with performance metrics and examples. Ensemble methods, deep learning networks, including MLP, DNNs, RNNs, CNNs, and transformers, are explored for their growing role in chemical applications. Practical workshops guide readers through predictive modeling using advanced ML algorithms. The chapter culminates with insights into science-guided ML, advocating for a hybrid approach that enhances model accuracy. The extensive bibliography offers resources for further research and practical implementation. This chapter aims to be a thorough primer on ML's practical application in chemical engineering, particularly for polyolefin production, and sets the stage for continued learning in subsequent chapters. Please cite the original work [169,170] when referencing.
Wavelet-Guided Acceleration of Text Inversion in Diffusion-Based Image Editing
Authors: Authors: Gwanhyeong Koo, Sunjae Yoon, Chang D. Yoo
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
In the field of image editing, Null-text Inversion (NTI) enables fine-grained editing while preserving the structure of the original image by optimizing null embeddings during the DDIM sampling process. However, the NTI process is time-consuming, taking more than two minutes per image. To address this, we introduce an innovative method that maintains the principles of the NTI while accelerating the image editing process. We propose the WaveOpt-Estimator, which determines the text optimization endpoint based on frequency characteristics. Utilizing wavelet transform analysis to identify the image's frequency characteristics, we can limit text optimization to specific timesteps during the DDIM sampling process. By adopting the Negative-Prompt Inversion (NPI) concept, a target prompt representing the original image serves as the initial text value for optimization. This approach maintains performance comparable to NTI while reducing the average editing time by over 80% compared to the NTI method. Our method presents a promising approach for efficient, high-quality image editing based on diffusion models.
Power System Fault Diagnosis with Quantum Computing and Efficient Gate Decomposition
Abstract
Power system fault diagnosis is crucial for identifying the location and causes of faults and providing decision-making support for power dispatchers. However, most classical methods suffer from significant time-consuming, memory overhead, and computational complexity issues as the scale of the power system concerned increases. With rapid development of quantum computing technology, the combinatorial optimization method based on quantum computing has shown certain advantages in computational time over existing methods. Given this background, this paper proposes a quantum computing based power system fault diagnosis method with the Quantum Approximate Optimization Algorithm (QAOA). The proposed method reformulates the fault diagnosis problem as a Hamiltonian by using Ising model, which completely preserves the coupling relationship between faulty components and various operations of protective relays and circuit breakers. Additionally, to enhance problem-solving efficiency under current equipment limitations, the symmetric equivalent decomposition method of multi-z-rotation gate is proposed. Furthermore, the small probability characteristics of power system events is utilized to reduce the number of qubits. Simulation results based on the test system show that the proposed methods can achieve the same optimal results with a faster speed compared with the classical higher-order solver provided by D-Wave.
Clickbait vs. Quality: How Engagement-Based Optimization Shapes the Content Landscape in Online Platforms
Abstract
Online content platforms commonly use engagement-based optimization when making recommendations. This encourages content creators to invest in quality, but also rewards gaming tricks such as clickbait. To understand the total impact on the content landscape, we study a game between content creators competing on the basis of engagement metrics and analyze the equilibrium decisions about investment in quality and gaming. First, we show the content created at equilibrium exhibits a positive correlation between quality and gaming, and we empirically validate this finding on a Twitter dataset. Using the equilibrium structure of the content landscape, we then examine the downstream performance of engagement-based optimization along several axes. Perhaps counterintuitively, the average quality of content consumed by users can decrease at equilibrium as gaming tricks become more costly for content creators to employ. Moreover, engagement-based optimization can perform worse in terms of user utility than a baseline with random recommendations, and engagement-based optimization is also suboptimal in terms of realized engagement relative to quality-based optimization. Altogether, our results highlight the need to consider content creator incentives when evaluating a platform's choice of optimization metric.
EDAF: An End-to-End Delay Analytics Framework for 5G-and-Beyond Networks
Authors: Authors: Samie Mostafavi, Marius Tillner, Gourav Prateek Sharma, James Gross
Subjects: Networking and Internet Architecture (cs.NI)
Abstract
Supporting applications in emerging domains like cyber-physical systems and human-in-the-loop scenarios typically requires adherence to strict end-to-end delay guarantees. Contributions of many tandem processes unfolding layer by layer within the wireless network result in violations of delay constraints, thereby severely degrading application performance. Meeting the application's stringent requirements necessitates coordinated optimization of the end-to-end delay by fine-tuning all contributing processes. To achieve this task, we designed and implemented EDAF, a framework to decompose packets' end-to-end delays and determine each component's significance for 5G network. We showcase EDAF on OpenAirInterface 5G uplink, modified to report timestamps across the data plane. By applying the obtained insights, we optimized end-to-end uplink delay by eliminating segmentation and frame-alignment delays, decreasing average delay from 12ms to 4ms.
Improving the Accuracy of Analog-Based In-Memory Computing Accelerators Post-Training
Authors: Authors: Corey Lammie, Athanasios Vasilopoulos, Julian Büchel, Giacomo Camposampiero, Manuel Le Gallo, Malte Rasch, Abu Sebastian
Abstract
Analog-Based In-Memory Computing (AIMC) inference accelerators can be used to efficiently execute Deep Neural Network (DNN) inference workloads. However, to mitigate accuracy losses, due to circuit and device non-idealities, Hardware-Aware (HWA) training methodologies must be employed. These typically require significant information about the underlying hardware. In this paper, we propose two Post-Training (PT) optimization methods to improve accuracy after training is performed. For each crossbar, the first optimizes the conductance range of each column, and the second optimizes the input, i.e, Digital-to-Analog Converter (DAC), range. It is demonstrated that, when these methods are employed, the complexity during training, and the amount of information about the underlying hardware can be reduced, with no notable change in accuracy ($\leq$0.1%) when finetuning the pretrained RoBERTa transformer model for all General Language Understanding Evaluation (GLUE) benchmark tasks. Additionally, it is demonstrated that further optimizing learned parameters PT improves accuracy.
Evolutionary Multi-Objective Optimization of Large Language Model Prompts for Balancing Sentiments
Authors: Authors: Jill Baumann, Oliver Kramer
Subjects: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Abstract
The advent of large language models (LLMs) such as ChatGPT has attracted considerable attention in various domains due to their remarkable performance and versatility. As the use of these models continues to grow, the importance of effective prompt engineering has come to the fore. Prompt optimization emerges as a crucial challenge, as it has a direct impact on model performance and the extraction of relevant information. Recently, evolutionary algorithms (EAs) have shown promise in addressing this issue, paving the way for novel optimization strategies. In this work, we propose a evolutionary multi-objective (EMO) approach specifically tailored for prompt optimization called EMO-Prompts, using sentiment analysis as a case study. We use sentiment analysis capabilities as our experimental targets. Our results demonstrate that EMO-Prompts effectively generates prompts capable of guiding the LLM to produce texts embodying two conflicting emotions simultaneously.
Multiobjective Optimization Analysis for Finding Infrastructure-as-Code Deployment Configurations
Authors: Authors: Eneko Osaba, Josu Diaz-de-Arcaya, Juncal Alonso, Jesus L. Lobo, Gorka Benguria, Iñaki Etxaniz
Subjects: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI)
Abstract
Multiobjective optimization is a hot topic in the artificial intelligence and operations research communities. The design and development of multiobjective methods is a frequent task for researchers and practitioners. As a result of this vibrant activity, a myriad of techniques have been proposed in the literature to date, demonstrating a significant effectiveness for dealing with situations coming from a wide range of real-world areas. This paper is focused on a multiobjective problem related to optimizing Infrastructure-as-Code deployment configurations. The system implemented for solving this problem has been coined as IaC Optimizer Platform (IOP). Despite the fact that a prototypical version of the IOP has been introduced in the literature before, a deeper analysis focused on the resolution of the problem is needed, in order to determine which is the most appropriate multiobjective method for embedding in the IOP. The main motivation behind the analysis conducted in this work is to enhance the IOP performance as much as possible. This is a crucial aspect of this system, deeming that it will be deployed in a real environment, as it is being developed as part of a H2020 European project. Going deeper, we resort in this paper to nine different evolutionary computation-based multiobjective algorithms. For assessing the quality of the considered solvers, 12 different problem instances have been generated based on real-world settings. Results obtained by each method after 10 independent runs have been compared using Friedman's non-parametric tests. Findings reached from the tests carried out lad to the creation of a multi-algorithm system, capable of applying different techniques according to the user's needs.
BPDO:Boundary Points Dynamic Optimization for Arbitrary Shape Scene Text Detection
Abstract
Arbitrary shape scene text detection is of great importance in scene understanding tasks. Due to the complexity and diversity of text in natural scenes, existing scene text algorithms have limited accuracy for detecting arbitrary shape text. In this paper, we propose a novel arbitrary shape scene text detector through boundary points dynamic optimization(BPDO). The proposed model is designed with a text aware module (TAM) and a boundary point dynamic optimization module (DOM). Specifically, the model designs a text aware module based on segmentation to obtain boundary points describing the central region of the text by extracting a priori information about the text region. Then, based on the idea of deformable attention, it proposes a dynamic optimization model for boundary points, which gradually optimizes the exact position of the boundary points based on the information of the adjacent region of each boundary point. Experiments on CTW-1500, Total-Text, and MSRA-TD500 datasets show that the model proposed in this paper achieves a performance that is better than or comparable to the state-of-the-art algorithm, proving the effectiveness of the model.
Spintronic logic: from transducers to logic gates and circuits
Abstract
While magnetic solid-state memory has found commercial applications to date, magnetic logic has rather remained on a conceptual level so far. Here, we discuss open challenges of different spintronic logic approaches, which use magnetic excitations for computation. While different logic gate designs have been proposed and proof of concept experiments have been reported, no nontrivial operational spintronic circuit has been demonstrated due to many open challenges in spintronic circuit and system design. Furthermore, the integration of spintronic circuits in CMOS systems will require the usage of transducers between the electric (CMOS) and magnetic domains. We show that these transducers can limit the performance as well as the energy consumption of hybrid CMOS-spintronic systems. Hence, the optimization of transducer efficiency will be a major step towards competitive spintronic logic system.
Evolutionary Computation in the Era of Large Language Model: Survey and Roadmap
Authors: Authors: Xingyu Wu, Sheng-hao Wu, Jibin Wu, Liang Feng, Kay Chen Tan
Subjects: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Abstract
Large Language Models (LLMs), built upon Transformer-based architectures with massive pretraining on diverse data, have not only revolutionized natural language processing but also extended their prowess to various domains, marking a significant stride towards artificial general intelligence. The interplay between LLMs and Evolutionary Algorithms (EAs), despite differing in objectives and methodologies, reveals intriguing parallels, especially in their shared optimization nature, black-box characteristics, and proficiency in handling complex problems. Meanwhile, EA can not only provide an optimization framework for LLM's further enhancement under black-box settings but also empower LLM with flexible global search and iterative mechanism in applications. On the other hand, LLM's abundant domain knowledge enables EA to perform smarter searches, while its text processing capability assist in deploying EA across various tasks. Based on their complementary advantages, this paper presents a comprehensive review and forward-looking roadmap, categorizing their mutual inspiration into LLM-enhanced evolutionary optimization and EA-enhanced LLM. Some integrated synergy methods are further introduced to exemplify the amalgamation of LLMs and EAs in various application scenarios, including neural architecture search, code generation, software engineering, and text generation. As the first comprehensive review specifically focused on the EA research in the era of LLMs, this paper provides a foundational stepping stone for understanding and harnessing the collaborative potential of LLMs and EAs. By presenting a comprehensive review, categorization, and critical analysis, we contribute to the ongoing discourse on the cross-disciplinary study of these two powerful paradigms. The identified challenges and future directions offer guidance to unlock the full potential of this innovative collaboration.
Model-Assisted Learning for Adaptive Cooperative Perception of Connected Autonomous Vehicles
Abstract
Cooperative perception (CP) is a key technology to facilitate consistent and accurate situational awareness for connected and autonomous vehicles (CAVs). To tackle the network resource inefficiency issue in traditional broadcast-based CP, unicast-based CP has been proposed to associate CAV pairs for cooperative perception via vehicle-to-vehicle transmission. In this paper, we investigate unicast-based CP among CAV pairs. With the consideration of dynamic perception workloads and channel conditions due to vehicle mobility and dynamic radio resource availability, we propose an adaptive cooperative perception scheme for CAV pairs in a mixed-traffic autonomous driving scenario with both CAVs and human-driven vehicles. We aim to determine when to switch between cooperative perception and stand-alone perception for each CAV pair, and allocate communication and computing resources to cooperative CAV pairs for maximizing the computing efficiency gain under perception task delay requirements. A model-assisted multi-agent reinforcement learning (MARL) solution is developed, which integrates MARL for an adaptive CAV cooperation decision and an optimization model for communication and computing resource allocation. Simulation results demonstrate the effectiveness of the proposed scheme in achieving high computing efficiency gain, as compared with benchmark schemes.
SHINOBI: Shape and Illumination using Neural Object Decomposition via BRDF Optimization In-the-wild
Authors: Authors: Andreas Engelhardt, Amit Raj, Mark Boss, Yunzhi Zhang, Abhishek Kar, Yuanzhen Li, Deqing Sun, Ricardo Martin Brualla, Jonathan T. Barron, Hendrik P. A. Lensch, Varun Jampani
Subjects: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
Abstract
We present SHINOBI, an end-to-end framework for the reconstruction of shape, material, and illumination from object images captured with varying lighting, pose, and background. Inverse rendering of an object based on unconstrained image collections is a long-standing challenge in computer vision and graphics and requires a joint optimization over shape, radiance, and pose. We show that an implicit shape representation based on a multi-resolution hash encoding enables faster and robust shape reconstruction with joint camera alignment optimization that outperforms prior work. Further, to enable the editing of illumination and object reflectance (i.e. material) we jointly optimize BRDF and illumination together with the object's shape. Our method is class-agnostic and works on in-the-wild image collections of objects to produce relightable 3D assets for several use cases such as AR/VR, movies, games, etc. Project page: https://shinobi.aengelhardt.com Video: https://www.youtube.com/watch?v=iFENQ6AcYd8&feature=youtu.be
Fast Kronecker Matrix-Matrix Multiplication on GPUs
Authors: Authors: Abhinav Jangda, Mohit Yadav
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC)
Abstract
Kronecker Matrix-Matrix Multiplication (Kron-Matmul) is the multiplication of a matrix with the Kronecker Product of several smaller matrices. Kron-Matmul is a core operation for many scientific and machine learning computations. State-of-the-art Kron-Matmul implementations utilize existing tensor algebra operations, such as matrix multiplication, transpose, and tensor matrix multiplication. However, this design choice prevents several Kron-Matmul specific optimizations, thus, leaving significant performance on the table. To address this issue, we present FastKron, an efficient technique for Kron-Matmul on single and multiple GPUs. FastKron is independent of linear algebra operations enabling several new optimizations for Kron-Matmul. Thus, it performs up to 40.7x and 7.85x faster than existing implementations on 1 and 16 GPUs respectively.
Synchronization and Control of Chaotic Spur Gear System Using Type-II Fuzzy Controller Optimized via Whale Optimization Algorithm
Authors: Authors: Farnaz Rezay Morsagh, Mehdi Siahi, Soudabeh Soleymani
Abstract
Interval type-II Fuzzy Inference System (FIS) assumes a crucial role in determining the coefficients of the PID controller, thereby augmenting the controller's flexibility. Controlling chaotic systems presents inherent challenges and difficulties due to their sensitivity to initial conditions and the intricate dynamics that require precise and adaptive control strategies. This paper offers an exhaustive exploration into the coordination and regulation of a chaotic spur gear system, employing a Type-II Fuzzy Controller. The initial control parameters of the PID controller undergo optimization using the Whale Optimization Algorithm (WOA) to increase the overall system performance. The adaptability and strength of the suggested control system are tested in various scenarios, covering diverse reference inputs and uncertainties. The investigation comprehensively assesses the operational efficacy of the formulated controller, contrasting its performance with other methodologies. The outcomes highlight the impressive efficiency of the suggested strategy, confirming its supremacy in attaining synchronization and control within the turbulent spur gear system under demanding circumstances
Tailoring Semantic Communication at Network Edge: A Novel Approach Using Dynamic Knowledge Distillation
Abstract
Semantic Communication (SemCom) systems, empowered by deep learning (DL), represent a paradigm shift in data transmission. These systems prioritize the significance of content over sheer data volume. However, existing SemCom designs face challenges when applied to diverse computational capabilities and network conditions, particularly in time-sensitive applications. A key challenge is the assumption that diverse devices can uniformly benefit from a standard, large DL model in SemCom systems. This assumption becomes increasingly impractical, especially in high-speed, high-reliability applications such as industrial automation or critical healthcare. Therefore, this paper introduces a novel SemCom framework tailored for heterogeneous, resource-constrained edge devices and computation-intensive servers. Our approach employs dynamic knowledge distillation (KD) to customize semantic models for each device, balancing computational and communication constraints while ensuring Quality of Service (QoS). We formulate an optimization problem and develop an adaptive algorithm that iteratively refines semantic knowledge on edge devices, resulting in better models tailored to their resource profiles. This algorithm strategically adjusts the granularity of distilled knowledge, enabling devices to maintain high semantic accuracy for precise inference tasks, even under unstable network conditions. Extensive simulations demonstrate that our approach significantly reduces model complexity for edge devices, leading to better semantic extraction and achieving the desired QoS.
Keyword: adam
There is no result
Keyword: gradient
Bilevel Optimization under Unbounded Smoothness: A New Algorithm and Convergence Analysis
Authors: Authors: Jie Hao, Xiaochuan Gong, Mingrui Liu
Subjects: Machine Learning (cs.LG); Optimization and Control (math.OC)
Abstract
Bilevel optimization is an important formulation for many machine learning problems. Current bilevel optimization algorithms assume that the gradient of the upper-level function is Lipschitz. However, recent studies reveal that certain neural networks such as recurrent neural networks (RNNs) and long-short-term memory networks (LSTMs) exhibit potential unbounded smoothness, rendering conventional bilevel optimization algorithms unsuitable. In this paper, we design a new bilevel optimization algorithm, namely BO-REP, to address this challenge. This algorithm updates the upper-level variable using normalized momentum and incorporates two novel techniques for updating the lower-level variable: \textit{initialization refinement} and \textit{periodic updates}. Specifically, once the upper-level variable is initialized, a subroutine is invoked to obtain a refined estimate of the corresponding optimal lower-level variable, and the lower-level variable is updated only after every specific period instead of each iteration. When the upper-level problem is nonconvex and unbounded smooth, and the lower-level problem is strongly convex, we prove that our algorithm requires $\widetilde{\mathcal{O}}(1/\epsilon^4)$ iterations to find an $\epsilon$-stationary point in the stochastic setting, where each iteration involves calling a stochastic gradient or Hessian-vector product oracle. Notably, this result matches the state-of-the-art complexity results under the bounded smoothness setting and without mean-squared smoothness of the stochastic gradient, up to logarithmic factors. Our proof relies on novel technical lemmas for the periodically updated lower-level variable, which are of independent interest. Our experiments on hyper-representation learning, hyperparameter optimization, and data hyper-cleaning for text classification tasks demonstrate the effectiveness of our proposed algorithm.
Transient dynamics under structured perturbations: bridging unstructured and structured pseudospectra
Authors: Authors: Christian Lubich, Nicola Guglielmi
Abstract
The structured $\varepsilon$-stability radius is introduced as a quantity to assess the robustness of transient bounds of solutions to linear differential equations under structured perturbations of the matrix. This applies to general linear structures such as complex or real matrices with a given sparsity pattern or with restricted range and corange, or special classes such as Toeplitz matrices. The notion conceptually combines unstructured and structured pseudospectra in a joint pseudospectrum, allowing for the use of resolvent bounds as with unstructured pseudospectra and for structured perturbations as with structured pseudospectra. We propose and study an algorithm for computing the structured $\varepsilon$-stability radius. This algorithm solves eigenvalue optimization problems via suitably discretized rank-1 matrix differential equations that originate from a gradient system. The proposed algorithm has essentially the same computational cost as the known rank-1 algorithms for computing unstructured and structured stability radii. Numerical experiments illustrate the behavior of the algorithm.
Convex and Bilevel Optimization for Neuro-Symbolic Inference and Learning
Authors: Authors: Charles Dickens, Changyu Gao, Connor Pryor, Stephen Wright, Lise Getoor
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Optimization and Control (math.OC)
Abstract
We address a key challenge for neuro-symbolic (NeSy) systems by leveraging convex and bilevel optimization techniques to develop a general gradient-based framework for end-to-end neural and symbolic parameter learning. The applicability of our framework is demonstrated with NeuPSL, a state-of-the-art NeSy architecture. To achieve this, we propose a smooth primal and dual formulation of NeuPSL inference and show learning gradients are functions of the optimal dual variables. Additionally, we develop a dual block coordinate descent algorithm for the new formulation that naturally exploits warm-starts. This leads to over 100x learning runtime improvements over the current best NeuPSL inference method. Finally, we provide extensive empirical evaluations across $8$ datasets covering a range of tasks and demonstrate our learning framework achieves up to a 16% point prediction performance improvement over alternative learning methods.
Enabling On-device Continual Learning with Binary Neural Networks
Authors: Authors: Lorenzo Vorabbi, Davide Maltoni, Guido Borghi, Stefano Santi
Abstract
On-device learning remains a formidable challenge, especially when dealing with resource-constrained devices that have limited computational capabilities. This challenge is primarily rooted in two key issues: first, the memory available on embedded devices is typically insufficient to accommodate the memory-intensive back-propagation algorithm, which often relies on floating-point precision. Second, the development of learning algorithms on models with extreme quantization levels, such as Binary Neural Networks (BNNs), is critical due to the drastic reduction in bit representation. In this study, we propose a solution that combines recent advancements in the field of Continual Learning (CL) and Binary Neural Networks to enable on-device training while maintaining competitive performance. Specifically, our approach leverages binary latent replay (LR) activations and a novel quantization scheme that significantly reduces the number of bits required for gradient computation. The experimental validation demonstrates a significant accuracy improvement in combination with a noticeable reduction in memory requirement, confirming the suitability of our approach in expanding the practical applications of deep learning in real-world scenarios.
Abstract
Heterogeneous Graph Neural Networks (HGNNs) are increasingly recognized for their performance in areas like the web and e-commerce, where resilience against adversarial attacks is crucial. However, existing adversarial attack methods, which are primarily designed for homogeneous graphs, fall short when applied to HGNNs due to their limited ability to address the structural and semantic complexity of HGNNs. This paper introduces HGAttack, the first dedicated gray box evasion attack method for heterogeneous graphs. We design a novel surrogate model to closely resemble the behaviors of the target HGNN and utilize gradient-based methods for perturbation generation. Specifically, the proposed surrogate model effectively leverages heterogeneous information by extracting meta-path induced subgraphs and applying GNNs to learn node embeddings with distinct semantics from each subgraph. This approach improves the transferability of generated attacks on the target HGNN and significantly reduces memory costs. For perturbation generation, we introduce a semantics-aware mechanism that leverages subgraph gradient information to autonomously identify vulnerable edges across a wide range of relations within a constrained perturbation budget. We validate HGAttack's efficacy with comprehensive experiments on three datasets, providing empirical analyses of its generated perturbations. Outperforming baseline methods, HGAttack demonstrated significant efficacy in diminishing the performance of target HGNN models, affirming the effectiveness of our approach in evaluating the robustness of HGNNs against adversarial attacks.
SymbolNet: Neural Symbolic Regression with Adaptive Dynamic Pruning
Authors: Authors: Ho Fung Tsoi, Vladimir Loncar, Sridhara Dasu, Philip Harris
Subjects: Machine Learning (cs.LG); High Energy Physics - Experiment (hep-ex); Instrumentation and Detectors (physics.ins-det)
Abstract
Contrary to the use of genetic programming, the neural network approach to symbolic regression can scale well with high input dimension and leverage gradient methods for faster equation searching. Common ways of constraining expression complexity have relied on multistage pruning methods with fine-tuning, but these often lead to significant performance loss. In this work, we propose SymbolNet, a neural network approach to symbolic regression in a novel framework that enables dynamic pruning of model weights, input features, and mathematical operators in a single training, where both training loss and expression complexity are optimized simultaneously. We introduce a sparsity regularization term per pruning type, which can adaptively adjust its own strength and lead to convergence to a target sparsity level. In contrast to most existing symbolic regression methods that cannot efficiently handle datasets with more than $O$(10) inputs, we demonstrate the effectiveness of our model on the LHC jet tagging task (16 inputs), MNIST (784 inputs), and SVHN (3072 inputs).
Better Explain Transformers by Illuminating Important Information
Authors: Authors: Linxin Song, Yan Cui, Ao Luo, Freddy Lecue, Irene Li
Abstract
Transformer-based models excel in various natural language processing (NLP) tasks, attracting countless efforts to explain their inner workings. Prior methods explain Transformers by focusing on the raw gradient and attention as token attribution scores, where non-relevant information is often considered during explanation computation, resulting in confusing results. In this work, we propose highlighting the important information and eliminating irrelevant information by a refined information flow on top of the layer-wise relevance propagation (LRP) method. Specifically, we consider identifying syntactic and positional heads as important attention heads and focus on the relevance obtained from these important heads. Experimental results demonstrate that irrelevant information does distort output attribution scores and then should be masked during explanation computation. Compared to eight baselines on both classification and question-answering datasets, our method consistently outperforms with over 3\% to 33\% improvement on explanation metrics, providing superior explanation performance. Our anonymous code repository is available at: https://github.com/LinxinS97/Mask-LRP
CLIP feature-based randomized control using images and text for multiple tasks and robots
Abstract
This study presents a control framework leveraging vision language models (VLMs) for multiple tasks and robots. Notably, existing control methods using VLMs have achieved high performance in various tasks and robots in the training environment. However, these methods incur high costs for learning control policies for tasks and robots other than those in the training environment. Considering the application of industrial and household robots, learning in novel environments where robots are introduced is challenging. To address this issue, we propose a control framework that does not require learning control policies. Our framework combines the vision-language CLIP model with a randomized control. CLIP computes the similarity between images and texts by embedding them in the feature space. This study employs CLIP to compute the similarity between camera images and text representing the target state. In our method, the robot is controlled by a randomized controller that simultaneously explores and increases the similarity gradients. Moreover, we fine-tune the CLIP to improve the performance of the proposed method. Consequently, we confirm the effectiveness of our approach through a multitask simulation and a real robot experiment using a two-wheeled robot and robot arm.
Cross-Modality Perturbation Synergy Attack for Person Re-identification
Authors: Authors: Yunpeng Gong, others
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
In recent years, there has been significant research focusing on addressing security concerns in single-modal person re-identification (ReID) systems that are based on RGB images. However, the safety of cross-modality scenarios, which are more commonly encountered in practical applications involving images captured by infrared cameras, has not received adequate attention. The main challenge in cross-modality ReID lies in effectively dealing with visual differences between different modalities. For instance, infrared images are typically grayscale, unlike visible images that contain color information. Existing attack methods have primarily focused on the characteristics of the visible image modality, overlooking the features of other modalities and the variations in data distribution among different modalities. This oversight can potentially undermine the effectiveness of these methods in image retrieval across diverse modalities. This study represents the first exploration into the security of cross-modality ReID models and proposes a universal perturbation attack specifically designed for cross-modality ReID. This attack optimizes perturbations by leveraging gradients from diverse modality data, thereby disrupting the discriminator and reinforcing the differences between modalities. We conducted experiments on two widely used cross-modality datasets, namely RegDB and SYSU, which not only demonstrated the effectiveness of our method but also provided insights for future enhancements in the robustness of cross-modality ReID systems.
Residual Based Error Estimator for Chemical-Mechanically Coupled Battery Active Particles
Authors: Authors: Raphael Schoof, Lennart Flür, Florian Tuschner, Willy Dörfler
Abstract
Adaptive finite element methods are a powerful tool to obtain numerical simulation results in a reasonable time. Due to complex chemical and mechanical couplings in lithium-ion batteries, numerical simulations are very helpful to investigate promising new battery active materials such as amorphous silicon featuring a higher energy density than graphite. Based on a thermodynamically consistent continuum model with large deformation and chemo-mechanically coupled approach, we compare three different spatial adaptive refinement strategies: Kelly-, gradient recovery- and residual based error estimation. For the residual based case, the strong formulation of the residual is explicitly derived. With amorphous silicon as example material, we investigate two 3D representative host particle geometries, reduced with symmetry assumptions to a 1D unit interval and a 2D elliptical domain. Our numerical studies show that the Kelly estimator overestimates the error, whereas the gradient recovery estimator leads to lower refinement levels and a good capture of the change of the lithium flux. The residual based error estimator reveals a strong dependency on the cell error part which can be improved by a more suitable choice of constants to be more efficient. In a 2D domain, the concentration has a larger influence on the mesh distribution than the Cauchy stress.
Keyword: sgd
There is no result
Keyword: optimization
IPR-NeRF: Ownership Verification meets Neural Radiance Field
Aligning Large Language Models with Counterfactual DPO
Bilevel Optimization under Unbounded Smoothness: A New Algorithm and Convergence Analysis
Transient dynamics under structured perturbations: bridging unstructured and structured pseudospectra
SMOOTHIE: A Theory of Hyper-parameter Optimization for Software Analytics
smooth''. This theory is useful since the influence onsmoothness'' of different hyper-parameter choices can be tested very quickly (e.g. for a deep learner, after just one epoch). To test this theory, this paper implements and tests SMOOTHIE, a novel hyper-parameter optimizer that guides its optimizations via considerations ofsmothness''. The experiments of this paper test SMOOTHIE on numerous SE tasks including (a) GitHub issue lifetime prediction; (b) detecting false alarms in static code warnings; (c) defect prediction, and (d) a set of standard ML datasets. In all these experiments, SMOOTHIE out-performed state-of-the-art optimizers. Better yet, SMOOTHIE ran 300% faster than the prior state-of-the art. We hence conclude that this theory (that hyper-parameter optimization is best viewed as asmoothing'' function for the decision landscape), is both theoretically interesting and practically very useful. To support open science and other researchers working in this area, all our scripts and datasets are available on-line at https://github.com/yrahul3910/smoothness-hpo/.Multiple Locally Linear Kernel Machines
Convex and Bilevel Optimization for Neuro-Symbolic Inference and Learning
QoS-Aware 3D Coverage Deployment of UAVs for Internet of Vehicles in Intelligent Transportation
Joint Beam Direction Control and Radio Resource Allocation in Dynamic Multi-beam LEO Satellite Networks
Measuring the Discrepancy between 3D Geometric Models using Directional Distance Fields
Applications of Machine Learning to Optimizing Polyolefin Manufacturing
Wavelet-Guided Acceleration of Text Inversion in Diffusion-Based Image Editing
Power System Fault Diagnosis with Quantum Computing and Efficient Gate Decomposition
Clickbait vs. Quality: How Engagement-Based Optimization Shapes the Content Landscape in Online Platforms
EDAF: An End-to-End Delay Analytics Framework for 5G-and-Beyond Networks
Improving the Accuracy of Analog-Based In-Memory Computing Accelerators Post-Training
Evolutionary Multi-Objective Optimization of Large Language Model Prompts for Balancing Sentiments
Multiobjective Optimization Analysis for Finding Infrastructure-as-Code Deployment Configurations
BPDO:Boundary Points Dynamic Optimization for Arbitrary Shape Scene Text Detection
Spintronic logic: from transducers to logic gates and circuits
Evolutionary Computation in the Era of Large Language Model: Survey and Roadmap
Model-Assisted Learning for Adaptive Cooperative Perception of Connected Autonomous Vehicles
SHINOBI: Shape and Illumination using Neural Object Decomposition via BRDF Optimization In-the-wild
Fast Kronecker Matrix-Matrix Multiplication on GPUs
Synchronization and Control of Chaotic Spur Gear System Using Type-II Fuzzy Controller Optimized via Whale Optimization Algorithm
Tailoring Semantic Communication at Network Edge: A Novel Approach Using Dynamic Knowledge Distillation
Keyword: adam
There is no result
Keyword: gradient
Bilevel Optimization under Unbounded Smoothness: A New Algorithm and Convergence Analysis
Transient dynamics under structured perturbations: bridging unstructured and structured pseudospectra
Convex and Bilevel Optimization for Neuro-Symbolic Inference and Learning
Enabling On-device Continual Learning with Binary Neural Networks
HGAttack: Transferable Heterogeneous Graph Adversarial Attack
SymbolNet: Neural Symbolic Regression with Adaptive Dynamic Pruning
Better Explain Transformers by Illuminating Important Information
CLIP feature-based randomized control using images and text for multiple tasks and robots
Cross-Modality Perturbation Synergy Attack for Person Re-identification
Residual Based Error Estimator for Chemical-Mechanically Coupled Battery Active Particles
Keyword: super-resolution
There is no result