Mitigating Asymmetric Nonlinear Weight Update Effects in Hardware Neural Network Based on Analog Resistive Synapse
檔案
本條目沒有附加檔案。
日期
2018-03-01
作者
類型
系所
期刊名稱
IEEE JOURNAL ON EMERGING AND SELECTED TOPICS IN CIRCUITS AND SYSTEMS
卷期與頁次
第 8 卷,頁 116–124
ISSN
2156-3357
語言
英文
摘要
Asymmetric nonlinear weight update is considered as one of the major obstacles for realizing hardware neural networks based on analog resistive synapses, because it significantly compromises the online training capability. This paper provides new solutions to this critical issue through co-optimization with the hardware-applicable deep-learning algorithms. New insights on engineering activation functions and a threshold weight update scheme effectively suppress the undesirable training noise induced by inaccurate weight update. We successfully trained a two-layer perceptron network online and improved the classification accuracy of MNIST handwritten digit data set to 87.8%/94.8% by using 6-/8-b analog synapses, respectively, with extremely high asymmetric nonlinearity.
所屬典藏
期刊論文引用此條目
Chang, C.-C., Chen, P.-C., Chou, T., Wang, I.-T., Hudec, B., Chang, C.-C., Tsai, C.-M., Chang, T.-S., & Hou, T.-H. (2018). Mitigating Asymmetric Nonlinear Weight Update Effects in Hardware Neural Network Based on Analog Resistive Synapse. IEEE JOURNAL ON EMERGING AND SELECTED TOPICS IN CIRCUITS AND SYSTEMS, 8, 116–124. https://doi.org/10.1109/JETCAS.2017.2771529
引用文字依 APA 第 7 版格式自動產生,使用前請再核對。
分享
永久網址:https://ir.lib.nycu.edu.tw/handle/11536/144798