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. |
Year | Venue | Field |
---|---|---|
2016 | arXiv: Neural and Evolutionary Computing | Convolutional code,Convolution,Efficient energy use,Computer science,Matrix (mathematics),Neuromorphic engineering,Algorithm,Theoretical computer science,Artificial intelligence,Deep learning,Machine learning,Binary number |
DocType | Volume | Citations |
Journal | abs/1606.02407 | 1 |
PageRank | References | Authors |
0.36 | 30 | 9 |
Name | Order | Citations | PageRank |
---|---|---|---|
Rathinakumar Appuswamy | 1 | 239 | 14.94 |
Tapan K Nayak | 2 | 21 | 3.93 |
John V. Arthur | 3 | 790 | 44.72 |
Steven K. Esser | 4 | 266 | 16.08 |
Paul Merolla | 5 | 513 | 27.57 |
Jeffrey L. McKinstry | 6 | 163 | 9.40 |
Timothy Melano | 7 | 121 | 5.21 |
Myron Flickner | 8 | 913 | 179.69 |
Dharmendra S. Modha | 9 | 3004 | 193.20 |