在云对象存储上通过近数据计算加速迁移学习
Accelerating Transfer Learning with Near-Data Computation on Cloud Object Stores
AI总结：
针对云存储解耦下迁移学习的网络瓶颈与资源效率问题，提出基于存储下推的HAPI系统，通过特征提取阶段计算划分和存储端批量自适应，实现最高2.5倍训练加速。
摘要：
存储解耦是当今云计算的基础，将部分计算下推到存储层可对其形成天然补充，从而缓解存储层与计算层之间潜在的网络瓶颈。我们聚焦于迁移学习（Transfer Learning，TL）——一种通过在相关任务上复用现有知识来普及机器学习的广泛应用技术——展示了存储下推如何为机器学习训练带来收益。我们提出了HAPI，这是一种新型迁移学习处理系统，核心是两种互补的技术，用于应对解耦带来的挑战。首先，应用必须在各层之间谨慎平衡执行以保障性能。HAPI在特征提取阶段明智地划分迁移学习计算，生成的下推操作不仅能优化网络耗时，还能通过跨层重叠连续训练迭代的执行，缩短迁移学习的总训练时间。其次，运营方希望存储端计算资源具备资源效率。HAPI采用存储端批量大小自适应机制，可在不影响训练精度的前提下提升存储端下推并发度。HAPI最高可实现2.5倍的训练加速，且在86.8%的情况下能选择性能最优的划分点，或选择与最优值差距不超过5%的划分点。
英文摘要：
Storage disaggregation underlies today's cloud and is naturally complemented by pushing down some computation to storage, thus mitigating the potential network bottleneck between the storage and compute tiers. We show how ML training benefits from storage pushdowns by focusing on transfer learning (TL), the widespread technique that democratizes ML by reusing existing knowledge on related tasks. We propose HAPI, a new TL processing system centered around two complementary techniques that address challenges introduced by disaggregation. First, applications must carefully balance execution across tiers for performance. HAPI judiciously splits the TL computation during the feature extraction phase yielding pushdowns that not only improve network time but also improve total TL training time by overlapping the execution of consecutive training iterations across tiers. Second, operators want resource efficiency from the storage-side computational resources. HAPI employs storage-side batch size adaptation allowing increased storage-side pushdown concurrency without affecting training accuracy. HAPI yields up to 2.5x training speed-up while choosing in 86.8% of cases the best performing split point or one that is at most 5% off from the best.
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。
