Abstract
A Self-rePAiring spiking Neural NEtwoRk(SPANNER) hardware architecture is presented in this paper. Itis based on a software model of an astrocyte-neuron networkwhich previously demonstrated the ability to self-detect faultsand self-repair autonomously. Experimental results in this papershow that when faults occur at the synapse, remaining healthysynapses of the same neuron are enhanced by the feedback fromthe astrocyte, which enables the system functionality to bemaintained. This is the first time that astrocytes cells mergedwithin spiking neurons demonstrate a self-repairing capability inhardware. This achieves a much more fine-grained repaircapability in hardware compared to the conventional faulttolerance techniques.
Original language | English |
---|---|
Title of host publication | Unknown Host Publication |
Publisher | IEEE |
Pages | 1350-1353 |
Number of pages | 4 |
ISBN (Print) | 978-1-4799-5340-0 |
DOIs | |
Publication status | Published online - 11 Aug 2016 |
Event | IEEE International Symposium on Circuits and Systems (ISCAS) - Canada Duration: 11 Aug 2016 → … |
Conference
Conference | IEEE International Symposium on Circuits and Systems (ISCAS) |
---|---|
Period | 11/08/16 → … |
Keywords
- Astrocyte
- spiking neural networks
- fault tolerance
- FPGAs
- repair