An On-Chip Learning Neuromorphic Autoencoder With Current-Mode Transposable Memory Read and Virtual Lookup Table
| Autor: | Hwasuk Cho, Hong-June Park, Kihwan Seong, Jae-Yoon Sim, Hyunwoo Son, Byungsub Kim |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
010302 applied physics
Spiking neural network Restricted Boltzmann machine Databases Factual Computer science 020208 electrical & electronic engineering Biomedical Engineering 02 engineering and technology 01 natural sciences Autoencoder Computational science Machine Learning Synaptic weight Neuromorphic engineering 0103 physical sciences Lookup table 0202 electrical engineering electronic engineering information engineering Image Processing Computer-Assisted Unsupervised learning Humans Neural Networks Computer Electrical and Electronic Engineering Facial Recognition MNIST database |
| Zdroj: | IEEE transactions on biomedical circuits and systems. 12(1) |
| ISSN: | 1940-9990 |
| Popis: | This paper presents an IC implementation of on-chip learning neuromorphic autoencoder unit in a form of rate-based spiking neural network. With a current-mode signaling scheme embedded in a 500 × 500 6b SRAM-based memory, the proposed architecture achieves simultaneous processing of multiplications and accumulations. In addition, a transposable memory read for both forward and backward propagations and a virtual lookup table are also proposed to perform an unsupervised learning of restricted Boltzmann machine. The IC is fabricated using 28-nm CMOS process and is verified in a three-layer network of encoder-decoder pair for training and recovery of images with two-dimensional 16 × 16 pixels. With a dataset of 50 digits, the IC shows a normalized root mean square error of 0.078. Measured energy efficiencies are 4.46 pJ per synaptic operation for inference and 19.26 pJ per synaptic weight update for learning, respectively. The learning performance is also estimated by simulations if the proposed hardware architecture is extended to apply to a batch training of 60 000 MNIST datasets. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|