arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2510.08522cs.LGcs.DC
收藏

DYNAMIX:分布式机器学习系统中基于强化学习的自适应批大小优化

DYNAMIX: RL-based Adaptive Batch Size Optimization in Distributed Machine Learning Systems

Case Western Reserve University(凯斯西储大学);Shanghai Jiao Tong University(上海交通大学)

Yuanjun Dai, Keqiang He, An Wang

AI总结:

DYNAMIX利用强化学习(PPO)自适应优化分布式训练中的批大小,无需系统建模,在多种环境下提升准确率最高6.3%,减少训练时间46%,并具备良好的可扩展性与策略迁移能力。

摘要:

现有的分布式机器学习中的批大小选择方法依赖于静态分配或简单启发式方法,无法适应异构、动态的计算环境。我们提出了DYNAMIX,一个强化学习框架,将批大小优化表述为使用近端策略优化(PPO)的序列决策问题。我们的方法采用多维状态表示,涵盖网络级指标、系统级资源利用率和训练统计效率指标,以便在多样化的计算资源中做出明智决策。我们的方法无需显式系统建模,同时能无缝集成到现有的分布式训练框架中。通过在多样化工作负载、硬件配置和网络条件下的评估,DYNAMIX在最终模型准确率上实现了高达6.3%的提升,并将总训练时间减少了46%。我们的可扩展性实验表明,当集群规模增加到32个节点时,DYNAMIX保持了最佳性能,而策略迁移实验表明,学习到的策略能有效地泛化到相关的模型架构上。

英文摘要:

Existing batch size selection approaches in distributed machine learning rely on static allocation or simplistic heuristics that fail to adapt to heterogeneous, dynamic computing environments. We present DYNAMIX, a reinforcement learning framework that formulates batch size optimization as a sequential decision-making problem using Proximal Policy Optimization (PPO). Our approach employs a multi-dimensional state representation encompassing network-level metrics, system-level resource utilization, and training statistical efficiency indicators to enable informed decision-making across diverse computational resources. Our approach eliminates the need for explicit system modeling while integrating seamlessly with existing distributed training frameworks. Through evaluations across diverse workloads, hardware configurations, and network conditions, DYNAMIX achieves up to 6.3% improvement in the final model accuracy and 46% reduction in the total training time. Our scalability experiments demonstrate that DYNAMIX maintains the best performance as cluster size increases to 32 nodes, while policy transfer experiments show that learned policies generalize effectively across related model architectures.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑