Delay learning architectures for memory and classification

NEUROCOMPUTING

卷:138

Article

Hussain, Shaista¹  Basu, Arindam¹  Wang, Runchun Mark²  Hamilton, Tara Julia^2,3 

[1] Nanyang Technol Univ, Sch Elect & Elect Engn, Singapore 639798, Singapore

[2] Univ Western Sydney, Penrith, NSW 2751, Australia

[3] Univ New S Wales, Sch Elect Engn & Telecommun, Sydney, NSW 2052, Australia

关键词: Neuromorphic;    Spiking neural networks;    Delay-based learning;   

DOI  :  10.1016/j.neucom.2013.09.052

来源: Elsevier

【 摘 要 】

We present a neuromorphic spiking neural network, the DELTRON, that can remember and store patterns by changing the delays of every connection as opposed to modifying the weights. The advantage of this architecture over traditional weight-based ones is simpler hardware implementation without multipliers or digital-analog converters (DACs) as well as being suited to time-based computing. The name is derived due to similarity in the learning rule with an earlier architecture called tempotron. The DELTRON can remember more patterns than other delay-based networks by modifying a few delays to remember the most 'salient' or synchronous part of every spike pattern. We present simulations of memory capacity and classification ability of the DELTRON for different random spatio-temporal spike patterns. The memory capacity for noisy spike patterns and missing spikes is also shown. Finally, we present SPICE simulation results of the core circuits involved in a reconfigurable mixed signal implementation of this architecture. (C) 2014 Elsevier B.V. All rights reserved.

【 授权许可】

Free