Approximate Programming Design for Enhancing Energy, Endurance and Performance of Neural Network Training on NVM-based Systems

Chien Chung Ho, Wei Chen Wang, Te Hao Hsu, Zhi Duan Jiang, Yung Chun Li

研究成果: Conference contribution

1 引文 斯高帕斯(Scopus)

摘要

Recently, it is found non-volatile memories (NVMs) offer opportunities for mitigating issues of neural network training on DRAM-based systems by taking advantage of its near-zero leakage power and high scalability properties. However, it brings the new challenges on energy consumption, lifetime and performance degradation caused by the massive weight/bias updates performed during training phases. To tackle these issues, this work proposes an approximate write-once memory (WOM) code method with considering the characteristics of weight updates and error tolerability of NNs. In particular, the proposed method aims to effectively reduce the number of writes on NVMs. The experimental results demonstrate that great enhancement on energy consumption, endurance and write performance can be simultaneously achieved without sacrificing the inference accuracy.

原文English
主出版物標題Proceedings - 10th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2021
發行者Institute of Electrical and Electronics Engineers Inc.
ISBN(電子)9781665423755
DOIs
出版狀態Published - 2021
事件10th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2021 - Virtual, Online, China
持續時間: 2021 8月 182021 8月 19

出版系列

名字Proceedings - 10th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2021

Conference

Conference10th IEEE Non-Volatile Memory Systems and Applications Symposium, NVMSA 2021
國家/地區China
城市Virtual, Online
期間21-08-1821-08-19

All Science Journal Classification (ASJC) codes

  • 硬體和架構
  • 安全、風險、可靠性和品質

指紋

深入研究「Approximate Programming Design for Enhancing Energy, Endurance and Performance of Neural Network Training on NVM-based Systems」主題。共同形成了獨特的指紋。

引用此