PyTorchKR Papers

최신 논문

총 3,273편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2505.05099

Balancing Client Participation in Federated Learning Using AoI

Federated Learning (FL) offers a decentralized framework that preserves data privacy while enabling collaborative model training across distributed clients. However, FL faces significant challenges due to limited communication resources, statistical heterogeneity, and the need for balanced client participation. This paper proposes an Age of Information (AoI)-based client selection policy that addresses these challenges by minimizing load imbalance through controlled selection intervals. Our method employs a decentralized Markov scheduling policy, allowing clients to independently manage participation based on age-dependent selection probabilities, which balances client updates across training rounds with minimal central oversight. We provide a convergence proof for our method, demonstrating that it ensures stable and efficient model convergence. Specifically, we derive optimal parameters for the Markov selection model to achieve balanced and consistent client participation, highlighting the benefits of AoI in enhancing convergence stability. Through extensive simulations, we demonstrate that our AoI-based method, particularly the optimal Markov variant, improves convergence over the FedAvg selection approach across both IID and non-IID data settings by $7.5\%$ and up to $20\%$. Our findings underscore the effectiveness of AoI-based scheduling for scalable, fair, and efficient FL systems across diverse learning environments.

arXiv:2505.05047

Neural Pathways to Program Success: Hopfield Networks for PERT Analysis

Project and task scheduling under uncertainty remains a fundamental challenge in program and project management, where accurate estimation of task durations and dependencies is critical for delivering complex, multi project systems. The Program Evaluation and Review Technique provides a probabilistic framework to model task variability and critical paths. In this paper, the author presents a novel formulation of PERT scheduling as an energy minimization problem within a Hopfield neural network architecture. By mapping task start times and precedence constraints into a neural computation framework, the networks inherent optimization dynamics is exploited to approximate globally consistent schedules. The author addresses key theoretical issues related to energy function differentiability, constraint encoding, and convergence, and extends the Hopfield model for structured precedence graphs. Numerical simulations on synthetic project networks comprising up to 1000 tasks demonstrate the viability of this approach, achieving near optimal makespans with minimal constraint violations. The findings suggest that neural optimization models offer a promising direction for scalable and adaptive project tasks scheduling under uncertainty in areas such as the agentic AI workflows, microservice based applications that the modern AI systems are being built upon.

arXiv:2504.16918

OptimAI: Optimization from Natural Language Using LLM-Powered AI Agents

Optimization plays a vital role in scientific research and practical applications. However, formulating a concrete optimization problem described in natural language into a mathematical form and selecting a suitable solver to solve the problem requires substantial domain expertise. We introduce OptimAI, a framework for solving Optimization problems described in natural language by leveraging LLM-powered AI agents, and achieve superior performance over current state-of-the-art methods. Our framework is built upon the following key roles: (1) a formulator that translates natural language problem descriptions into precise mathematical formulations; (2) a planner that constructs a high-level solution strategy prior to execution; and (3) a coder and a code critic capable of interacting with the environment and reflecting on outcomes to refine future actions. Ablation studies confirm that all roles are essential; removing the planner or code critic results in $5.8\times$ and $3.1\times$ drops in productivity, respectively. Furthermore, we introduce UCB-based debug scheduling to dynamically switch between alternative plans, yielding an additional $3.3\times$ productivity gain. Our design emphasizes multi-agent collaboration, and our experiments confirm that combining diverse models leads to performance gains. Our approach attains 88.1% accuracy on the NLP4LP dataset and 82.3% on the Optibench dataset, reducing error rates by 58% and 52%, respectively, over prior best results.

arXiv:2505.04241

Technology prediction of a 3D model using Neural Network

Accurate estimation of production times is critical for effective manufacturing scheduling, yet traditional methods relying on expert analysis or historical data often fall short in dynamic or customized production environments. This paper introduces a data-driven approach that predicts manufacturing steps and their durations directly from a product's 3D model. By rendering the model into multiple 2D images and leveraging a neural network inspired by the Generative Query Network, the method learns to map geometric features into time estimates for predefined production steps enabling scalable, adaptive, and precise process planning across varied product types.

arXiv:2505.04596

Dynamic Network Flow Optimization for Task Scheduling in PTZ Camera Surveillance Systems

This paper presents a novel approach for optimizing the scheduling and control of Pan-Tilt-Zoom (PTZ) cameras in dynamic surveillance environments. The proposed method integrates Kalman filters for motion prediction with a dynamic network flow model to enhance real-time video capture efficiency. By assigning Kalman filters to tracked objects, the system predicts future locations, enabling precise scheduling of camera tasks. This prediction-driven approach is formulated as a network flow optimization, ensuring scalability and adaptability to various surveillance scenarios. To further reduce redundant monitoring, we also incorporate group-tracking nodes, allowing multiple objects to be captured within a single camera focus when appropriate. In addition, a value-based system is introduced to prioritize camera actions, focusing on the timely capture of critical events. By adjusting the decay rates of these values over time, the system ensures prompt responses to tasks with imminent deadlines. Extensive simulations demonstrate that this approach improves coverage, reduces average wait times, and minimizes missed events compared to traditional master-slave camera systems. Overall, our method significantly enhances the efficiency, scalability, and effectiveness of surveillance systems, particularly in dynamic and crowded environments.

arXiv:2505.04354

Optimization Problem Solving Can Transition to Evolutionary Agentic Workflows

This position paper argues that optimization problem solving can transition from expert-dependent to evolutionary agentic workflows. Traditional optimization practices rely on human specialists for problem formulation, algorithm selection, and hyperparameter tuning, creating bottlenecks that impede industrial adoption of cutting-edge methods. We contend that an evolutionary agentic workflow, powered by foundation models and evolutionary search, can autonomously navigate the optimization space, comprising problem, formulation, algorithm, and hyperparameter spaces. Through case studies in cloud resource scheduling and ADMM parameter adaptation, we demonstrate how this approach can bridge the gap between academic innovation and industrial implementation. Our position challenges the status quo of human-centric optimization workflows and advocates for a more scalable, adaptive approach to solving real-world optimization problems.

arXiv:2505.04021

Prism: Unleashing GPU Sharing for Cost-Efficient Multi-LLM Serving

Serving large language models (LLMs) is expensive, especially for providers hosting many models, making cost reduction essential. The unique workload patterns of serving multiple LLMs (i.e., multi-LLM serving) create new opportunities and challenges for this task. The long-tail popularity of models and their long idle periods present opportunities to improve utilization through GPU sharing. However, existing GPU sharing systems lack the ability to adjust their resource allocation and sharing policies at runtime, making them ineffective at meeting latency service-level objectives (SLOs) under rapidly fluctuating workloads. This paper presents Prism, a multi-LLM serving system that unleashes the full potential of GPU sharing to achieve both cost efficiency and SLO attainment. At its core, Prism tackles a key limitation of existing systems$\unicode{x2014}$the lack of $\textit{cross-model memory coordination}$, which is essential for flexibly sharing GPU memory across models under dynamic workloads. Prism achieves this with two key designs. First, it supports on-demand memory allocation by dynamically mapping physical to virtual memory pages, allowing flexible memory redistribution among models that space- and time-share a GPU. Second, it improves memory efficiency through a two-level scheduling policy that dynamically adjusts sharing strategies based on models' runtime demands. Evaluations on real-world traces show that Prism achieves more than $2\times$ cost savings and $3.3\times$ SLO attainment compared to state-of-the-art systems.

arXiv:2505.03946

Decentralized Distributed Proximal Policy Optimization (DD-PPO) for High Performance Computing Scheduling on Multi-User Systems

Resource allocation in High Performance Computing (HPC) environments presents a complex and multifaceted challenge for job scheduling algorithms. Beyond the efficient allocation of system resources, schedulers must account for and optimize multiple performance metrics, including job wait time and system utilization. While traditional rule-based scheduling algorithms dominate the current deployments of HPC systems, the increasing heterogeneity and scale of those systems is expected to challenge the efficiency and flexibility of those algorithms in minimizing job wait time and maximizing utilization. Recent research efforts have focused on leveraging advancements in Reinforcement Learning (RL) to develop more adaptable and intelligent scheduling strategies. Recent RL-based scheduling approaches have explored a range of algorithms, from Deep Q-Networks (DQN) to Proximal Policy Optimization (PPO), and more recently, hybrid methods that integrate Graph Neural Networks with RL techniques. However, a common limitation across these methods is their reliance on relatively small datasets, and these methods face scalability issues when using large datasets. This study introduces a novel RL-based scheduler utilizing the Decentralized Distributed Proximal Policy Optimization (DD-PPO) algorithm, which supports large-scale distributed training across multiple workers without requiring parameter synchronization at every step. By eliminating reliance on centralized updates to a shared policy, the DD-PPO scheduler enhances scalability, training efficiency, and sample utilization. The validation dataset leveraged over 11.5 million real HPC job traces for comparing DD-PPO performance between traditional and advanced scheduling approaches, and the experimental results demonstrate improved scheduling performance in comparison to both rule-based schedulers and existing RL-based scheduling algorithms.

arXiv:2505.02714

Less is More: Efficient Weight Farcasting with 1-Layer Neural Network

Addressing the computational challenges inherent in training large-scale deep neural networks remains a critical endeavor in contemporary machine learning research. While previous efforts have focused on enhancing training efficiency through techniques such as gradient descent with momentum, learning rate scheduling, and weight regularization, the demand for further innovation continues to burgeon as model sizes keep expanding. In this study, we introduce a novel framework which diverges from conventional approaches by leveraging long-term time series forecasting techniques. Our method capitalizes solely on initial and final weight values, offering a streamlined alternative for complex model architectures. We also introduce a novel regularizer that is tailored to enhance the forecasting performance of our approach. Empirical evaluations conducted on synthetic weight sequences and real-world deep learning architectures, including the prominent large language model DistilBERT, demonstrate the superiority of our method in terms of forecasting accuracy and computational efficiency. Notably, our framework showcases improved performance while requiring minimal additional computational overhead, thus presenting a promising avenue for accelerating the training process across diverse tasks and architectures.

arXiv:2504.16351

Self-Nomination: Deep Learning for Decentralized CSI Feedback Reduction in MU-MIMO Systems

This paper introduces a novel deep learning-based user-side feedback reduction framework, termed self-nomination. The goal of self-nomination is to reduce the number of users (UEs) feeding back channel state information (CSI) to the base station (BS), by letting each UE decide whether to feed back based on its estimated likelihood of being scheduled and its potential contribution to precoding in a multiuser MIMO (MU-MIMO) downlink. Unlike SNR- or SINR-based thresholding methods, the proposed approach uses rich spatial channel statistics and learns nontrivial correlation effects that affect eventual MU-MIMO scheduling decisions. To train the self-nomination network under an average feedback constraint, we propose two different strategies: one based on direct optimization with gradient approximations, and another using policy gradient-based optimization with a stochastic Bernoulli policy to handle non-differentiable scheduling. The framework also supports proportional-fair scheduling by incorporating dynamic user weights. Numerical results confirm that the proposed self-nomination method significantly reduces CSI feedback overhead. Compared to baseline feedback methods, self-nomination can reduce feedback by as much as 65%, saving not only bandwidth but also allowing many UEs to avoid feedback altogether (and thus, potentially enter a sleep mode). Self-nomination achieves this significant savings with negligible reduction in sum-rate or fairness.

arXiv:2504.20108공식 코드 · Papers with Code

Swapped Logit Distillation via Bi-level Teacher Alignment

Knowledge distillation (KD) compresses the network capacity by transferring knowledge from a large (teacher) network to a smaller one (student). It has been mainstream that the teacher directly transfers knowledge to the student with its original distribution, which can possibly lead to incorrect predictions. In this article, we propose a logit-based distillation via swapped logit processing, namely Swapped Logit Distillation (SLD). SLD is proposed under two assumptions: (1) the wrong prediction occurs when the prediction label confidence is not the maximum; (2) the "natural" limit of probability remains uncertain as the best value addition to the target cannot be determined. To address these issues, we propose a swapped logit processing scheme. Through this approach, we find that the swap method can be effectively extended to teacher and student outputs, transforming into two teachers. We further introduce loss scheduling to boost the performance of two teachers' alignment. Extensive experiments on image classification tasks demonstrate that SLD consistently performs best among previous state-of-the-art methods.

arXiv:2504.18368

Renewable-Colocated Green Hydrogen Production: Optimal Scheduling and Profitability

We study the optimal green hydrogen production and energy market participation of a renewable-colocated hydrogen producer (RCHP) that utilizes onsite renewable generation for both hydrogen production and grid services. Under deterministic and stochastic profit-maximization frameworks, we analyze RCHP's multiple market participation models and derive closed-form optimal scheduling policies that dynamically allocate renewable energy to hydrogen production and electricity export to the wholesale market. Analytical characterizations of the RCHP's operating profit and the optimal sizing of renewable and electrolyzer capacities are obtained. We use real-time renewable production and electricity price data from three independent system operators to assess the impacts of hydrogen market prices, renewable generation, and electricity prices on RCHP's profitability.

arXiv:2504.17805

Fuzzy Logic -- Based Scheduling System for Part-Time Workforce

This paper explores the application of genetic fuzzy systems to efficiently generate schedules for a team of part-time student workers at a university. Given the preferred number of working hours and availability of employees, our model generates feasible solutions considering various factors, such as maximum weekly hours, required number of workers on duty, and the preferred number of working hours. The algorithm is trained and tested with availability data collected from students at the University of Cincinnati. The results demonstrate the algorithm's efficiency in producing schedules that meet operational criteria and its robustness in understaffed conditions.

arXiv:2504.15672

Comparative Analysis of Evolutionary Algorithms for Energy-Aware Production Scheduling

The energy transition is driving rapid growth in renewable energy generation, creating the need to balance energy supply and demand with energy price awareness. One such approach for manufacturers to balance their energy demand with available energy is energyaware production planning. Through energy-aware production planning, manufacturers can align their energy demand with dynamic grid conditions, supporting renewable energy integration while benefiting from lower prices and reduced emissions. Energy-aware production planning can be modeled as a multi-criteria scheduling problem, where the objectives extend beyond traditional metrics like makespan or required workers to also include minimizing energy costs and emissions. Due to market dynamics and the NP-hard multi-objective nature of the problem, evolutionary algorithms are widely used for energy-aware scheduling. However, existing research focuses on the design and analysis of single algorithms, with limited comparisons between different approaches. In this study, we adapt NSGA-III, HypE, and $\theta$-DEA as memetic metaheuristics for energy-aware scheduling to minimize makespan, energy costs, emissions, and the number of workers, within a real-time energy market context. These adapted metaheuristics present different approaches for environmental selection. In a comparative analysis, we explore differences in solution efficiency and quality across various scenarios which are based on benchmark instances from the literature and real-world energy market data. Additionally, we estimate upper bounds on the distance between objective values obtained with our memetic metaheuristics and reference sets obtained via an exact solver.

arXiv:2504.19835

Automated Generation of Precedence Graphs in Digital Value Chains for Automotive Production

This study examines the digital value chain in automotive manufacturing, focusing on the identification, software flashing, customization, and commissioning of electronic control units in vehicle networks. A novel precedence graph design is proposed to optimize this process chain using an automated scheduling algorithm, which combines structured data extraction from heterogeneous sources via natural language processing and classification techniques with mixed integer linear programming for efficient graph generation. The results show significant improvements in key metrics. The algorithm reduces the number of production stations equipped with expensive hardware and software to execute digital value chain processes, while also increasing capacity utilization through efficient scheduling and reduced idle time. Task parallelization is optimized, resulting in streamlined workflows and increased throughput. Compared to the traditional scheduling method, the automated approach has reduced preparation time by 50% and reduced scheduling activities, as it now takes two minutes to create the precedence graph. The flexibility of the algorithm's constraints allows for vehicle-specific configurations while maintaining high responsiveness, eliminating backup stations and facilitating the integration of new topologies. Automated scheduling significantly outperforms manual methods in efficiency, functionality, and adaptability.

arXiv:2504.15577

State-Aware IoT Scheduling Using Deep Q-Networks and Edge-Based Coordination

This paper addresses the challenge of energy efficiency management faced by intelligent IoT devices in complex application environments. A novel optimization method is proposed, combining Deep Q-Network (DQN) with an edge collaboration mechanism. The method builds a state-action-reward interaction model and introduces edge nodes as intermediaries for state aggregation and policy scheduling. This enables dynamic resource coordination and task allocation among multiple devices. During the modeling process, device status, task load, and network resources are jointly incorporated into the state space. The DQN is used to approximate and learn the optimal scheduling strategy. To enhance the model's ability to perceive inter-device relationships, a collaborative graph structure is introduced to model the multi-device environment and assist in decision optimization. Experiments are conducted using real-world IoT data collected from the FastBee platform. Several comparative and validation tests are performed, including energy efficiency comparisons across different scheduling strategies, robustness analysis under varying task loads, and evaluation of state dimension impacts on policy convergence speed. The results show that the proposed method outperforms existing baseline approaches in terms of average energy consumption, processing latency, and resource utilization. This confirms its effectiveness and practicality in intelligent IoT scenarios.

arXiv:2504.15465

LithOS: An Operating System for Efficient Machine Learning on GPUs

The surging demand for GPUs in datacenters for machine learning (ML) has made efficient GPU utilization crucial. However, meeting the diverse needs of ML models while optimizing resource usage is challenging. To enable transparent, fine-grained GPU management that maximizes utilization and energy efficiency while maintaining strong isolation, an operating system (OS) approach is needed. This paper introduces LithOS, a first step toward a GPU OS. LithOS includes the following new abstractions and mechanisms for efficient GPU resource management: (i) a novel TPC Scheduler that supports spatial scheduling at the granularity of individual TPCs, unlocking efficient TPC stealing between workloads; (ii) transparent kernel atomization to reduce head-of-line blocking and enable dynamic resource reallocation mid-execution; (iii) a lightweight hardware right-sizing mechanism that determines the minimal TPC resources needed per atom; and (iv) a transparent power management mechanism that reduces power consumption based on in-flight work behavior. We implement LithOS in Rust and evaluate its performance across extensive ML environments, comparing it to state-of-the-art solutions from NVIDIA and prior research. For inference stacking, LithOS reduces tail latencies by 13x compared to MPS; compared to the best SotA, it reduces tail latencies by 3x while improving aggregate throughput by 1.6x. In hybrid inference-training stacking, LithOS reduces tail latencies by 4.7x compared to MPS; compared to the best SotA, it reduces tail latencies 1.18x while improving aggregate throughput by 1.35x. Finally, for a modest performance hit under 4%, LithOS's right-sizing provides a quarter of GPU capacity savings on average, while for a 7% hit, its power management yields a quarter of a GPU's energy savings. Overall, LithOS increases GPU efficiency, establishing a foundation for future OS research on GPUs.

arXiv:2505.03323공식 코드 · Papers with Code

Unraveling the Rainbow: can value-based methods schedule?

Recently, deep reinforcement learning has emerged as a promising approach for solving complex combinatorial optimization problems. Broadly, deep reinforcement learning methods fall into two categories: policy-based and value-based. While value-based approaches have achieved notable success in domains such as the Arcade Learning Environment, the combinatorial optimization community has predominantly favored policy-based methods, often overlooking the potential of value-based algorithms. In this work, we conduct a comprehensive empirical evaluation of value-based algorithms, including the deep q-network and several of its advanced extensions, within the context of two complex combinatorial problems: the job-shop and the flexible job-shop scheduling problems, two fundamental challenges with multiple industrial applications. Our results challenge the assumption that policy-based methods are inherently superior for combinatorial optimization. We show that several value-based approaches can match or even outperform the widely adopted proximal policy optimization algorithm, suggesting that value-based strategies deserve greater attention from the combinatorial optimization community. Our code is openly available at: https://github.com/AJ-Correa/Unraveling-the-Rainbow.

arXiv:2505.00482

JointDiT: Enhancing RGB-Depth Joint Modeling with Diffusion Transformers

We present JointDiT, a diffusion transformer that models the joint distribution of RGB and depth. By leveraging the architectural benefit and outstanding image prior of the state-of-the-art diffusion transformer, JointDiT not only generates high-fidelity images but also produces geometrically plausible and accurate depth maps. This solid joint distribution modeling is achieved through two simple yet effective techniques that we propose, i.e., adaptive scheduling weights, which depend on the noise levels of each modality, and the unbalanced timestep sampling strategy. With these techniques, we train our model across all noise levels for each modality, enabling JointDiT to naturally handle various combinatorial generation tasks, including joint generation, depth estimation, and depth-conditioned image generation by simply controlling the timestep of each branch. JointDiT demonstrates outstanding joint generation performance. Furthermore, it achieves comparable results in depth estimation and depth-conditioned image generation, suggesting that joint distribution modeling can serve as a replaceable alternative to conditional generation. The project page is available at https://byungki-k.github.io/JointDiT/.

arXiv:2504.14725

Sensor Scheduling in Intrusion Detection Games with Uncertain Payoffs

We study the problem of sensor scheduling for an intrusion detection task. We model this as a two-player zero-sum game over a graph, where the defender (Player 1) seeks to identify the optimal strategy for scheduling sensor orientations to minimize the probability of missed detection at minimal cost, while the intruder (Player 2) aims to identify the optimal path selection strategy to maximize missed detection probability at minimal cost. The defender's strategy space grows exponentially with the number of sensors, making direct computation of the Nash Equilibrium (NE) strategies computationally expensive. To tackle this, we propose a distributed variant of the Weighted Majority algorithm that exploits the structure of the game's payoff matrix, enabling efficient computation of the NE strategies with provable convergence guarantees. Next, we consider a more challenging scenario where the defender lacks knowledge of the true sensor models and, consequently, the game's payoff matrix. For this setting, we develop online learning algorithms that leverage bandit feedback from sensors to estimate the NE strategies. By building on existing results from perturbation theory and online learning in matrix games, we derive high-probability order-optimal regret bounds for our algorithms. Finally, through simulations, we demonstrate the empirical performance of our proposed algorithms in both known and unknown payoff scenarios.