arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2107.09786cs.LG
收藏

使用异步训练的分裂学习的通信和计算减少

Communication and Computation Reduction for Split Learning using Asynchronous Training

Arizona State University(亚利桑那州立大学)

Xing Chen, Jingtao Li, Chaitali Chakrabarti

摘要:

分裂学习是一种很有前途的隐私保护分布式学习方案,它对边缘设备的计算要求低,但存在边缘设备与服务器之间通信开销大的缺点。为了减少通信开销,本文提出了一种基于丢失的异步训练方案,该方案更新客户端模型的频率较低,并且只在选定的时期发送/接收激活/梯度。为了进一步减少通信开销,在传输之前使用8位浮点对激活/梯度进行量化。所提出的通信减少方法的附加益处在于,由于客户端模型更新的数量减少,客户端侧的计算减少。此外,所提出的基于通信减少的分裂学习方法的隐私性几乎与传统的分裂学习相同。在CIFAR-10上对VGG 11、VGG 13和ResNet 18模型的仿真结果表明,在单客户端情况下,当精度下降小于0.5%时,通信开销减少了1.64倍-106.7倍,客户端计算量减少了2.86倍-32.1倍。对于5个和10个客户端的情况,通信成本降低是VGG 11的11.9倍和11.3倍,准确性损失为0.5%。

英文摘要:

Split learning is a promising privacy-preserving distributed learning scheme that has low computation requirement at the edge device but has the disadvantage of high communication overhead between edge device and server. To reduce the communication overhead, this paper proposes a loss-based asynchronous training scheme that updates the client-side model less frequently and only sends/receives activations/gradients in selected epochs. To further reduce the communication overhead, the activations/gradients are quantized using 8-bit floating point prior to transmission. An added benefit of the proposed communication reduction method is that the computations at the client side are reduced due to reduction in the number of client model updates. Furthermore, the privacy of the proposed communication reduction based split learning method is almost the same as traditional split learning. Simulation results on VGG11, VGG13 and ResNet18 models on CIFAR-10 show that the communication cost is reduced by 1.64x-106.7x and the computations in the client are reduced by 2.86x-32.1x when the accuracy degradation is less than 0.5% for the single-client case. For 5 and 10-client cases, the communication cost reduction is 11.9x and 11.3x on VGG11 for 0.5% loss in accuracy.

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。

↑