arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2208.11596cs.LGcs.ITmath.ITstat.ML
收藏

一种低复杂度的率失真优化可变比特率压缩方法

A Low-Complexity Approach to Rate-Distortion Optimized Variable Bit-Rate Compression for Split DNN Computing

Intel Labs(英特尔实验室)

Parual Datta, Nilesh Ahuja, V. Srinivasa Somayazulu, Omesh Tickoo

摘要:

拆分计算已经成为实现基于DNN的AI工作负载的最新范例,其中DNN模型被分成两部分,其中一部分在移动/客户端设备上执行,另一部分在边缘服务器(或云)上执行。将数据压缩应用于来自DNN的需要传输的中间张量,从而解决优化速率-准确度-复杂度权衡的挑战。现有的分裂计算方法采用基于ML的数据压缩,但需要针对不同的压缩级别重新训练整个DNN模型或其中的重要部分的参数。这会带来很高的计算和存储负担:从头开始训练完整的DNN模型在计算上要求很高,维护DNN参数的多个副本会增加存储需求,并且在推理期间切换完整的权重集会增加内存带宽。在本文中,我们提出了一种解决所有这些挑战的方法。它涉及瓶颈单元的系统设计和训练--简单、低成本的神经网络--可以在分裂点插入。我们的方法是非常轻量级的,无论是在训练和推理,非常有效,并实现了出色的率失真性能在一小部分的计算和存储开销相比,现有的方法。

英文摘要:

Split computing has emerged as a recent paradigm for implementation of DNN-based AI workloads, wherein a DNN model is split into two parts, one of which is executed on a mobile/client device and the other on an edge-server (or cloud). Data compression is applied to the intermediate tensor from the DNN that needs to be transmitted, addressing the challenge of optimizing the rate-accuracy-complexity trade-off. Existing split-computing approaches adopt ML-based data compression, but require that the parameters of either the entire DNN model, or a significant portion of it, be retrained for different compression levels. This incurs a high computational and storage burden: training a full DNN model from scratch is computationally demanding, maintaining multiple copies of the DNN parameters increases storage requirements, and switching the full set of weights during inference increases memory bandwidth. In this paper, we present an approach that addresses all these challenges. It involves the systematic design and training of bottleneck units - simple, low-cost neural networks - that can be inserted at the point of split. Our approach is remarkably lightweight, both during training and inference, highly effective and achieves excellent rate-distortion performance at a small fraction of the compute and storage overhead compared to existing methods.

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。

↑