Journal
IEEE JOURNAL ON EMERGING AND SELECTED TOPICS IN CIRCUITS AND SYSTEMS
Volume 8, Issue 1, Pages 116-124Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JETCAS.2017.2771529
Keywords
Neuromorphic computing; RRAM; synapse; multilayer perceptron
Categories
Funding
- Ministry of Science and Technology of Taiwan [MOST 105-2119-M-009-009]
- Research of Excellence Program [MOST 106-2633-E-009-001]
- Taiwan Semiconductor Manufacturing Company
- NCTU-UCB I-RiCE Program [MOST 106-2911-I-009-301]
Ask authors/readers for more resources
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.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available