Structured Convolution Matrices for Energy-efficient Deep learning
Neural and Evolutionary Computing
2016-06-09 v1 Artificial Intelligence
Computer Vision and Pattern Recognition
Machine Learning
Abstract
We derive a relationship between network representation in energy-efficient neuromorphic architectures and block Toplitz convolutional matrices. Inspired by this connection, we develop deep convolutional networks using a family of structured convolutional matrices and achieve state-of-the-art trade-off between energy efficiency and classification accuracy for well-known image recognition tasks. We also put forward a novel method to train binary convolutional networks by utilising an existing connection between noisy-rectified linear units and binary activations.
Cite
@article{arxiv.1606.02407,
title = {Structured Convolution Matrices for Energy-efficient Deep learning},
author = {Rathinakumar Appuswamy and Tapan Nayak and John Arthur and Steven Esser and Paul Merolla and Jeffrey Mckinstry and Timothy Melano and Myron Flickner and Dharmendra Modha},
journal= {arXiv preprint arXiv:1606.02407},
year = {2016}
}