Supervised training of spiking neural networks for robust deployment on mixed-signal neuromorphic processors
| Autor: | Dmitrii Zendrikov, Julian Büchel, Dylan R. Muir, Giacomo Indiveri, Sergio Solinas |
|---|---|
| Přispěvatelé: | University of Zurich, Muir, Dylan R |
| Rok vydání: | 2021 |
| Předmět: |
FOS: Computer and information sciences
Computer Science - Machine Learning Computer science Science Models Neurological Action Potentials Article Machine Learning (cs.LG) 03 medical and health sciences 0302 clinical medicine Biomimetics Robustness (computer science) Animals Humans 10194 Institute of Neuroinformatics 030304 developmental biology Neurons Spiking neural network Digital electronics 1000 Multidisciplinary 0303 health sciences Multidisciplinary Analogue electronics business.industry Supervised learning Mixed-signal integrated circuit Electrical and electronic engineering Neuromorphic engineering Computer engineering 570 Life sciences biology Medicine Neural Networks Computer Supervised Machine Learning Noise (video) business Algorithms 030217 neurology & neurosurgery |
| Zdroj: | Scientific Reports Scientific Reports, Vol 11, Iss 1, Pp 1-12 (2021) Scientific Reports, 11 (1) |
| ISSN: | 2045-2322 |
| DOI: | 10.1038/s41598-021-02779-x |
| Popis: | Mixed-signal analog/digital circuits emulate spiking neurons and synapses with extremely high energy efficiency, an approach known as "neuromorphic engineering". However, analog circuits are sensitive to process-induced variation among transistors in a chip ("device mismatch"). For neuromorphic implementation of Spiking Neural Networks (SNNs), mismatch causes parameter variation between identically-configured neurons and synapses. Each chip exhibits a different distribution of neural parameters, causing deployed networks to respond differently between chips. Current solutions to mitigate mismatch based on per-chip calibration or on-chip learning entail increased design complexity, area and cost, making deployment of neuromorphic devices expensive and difficult. Here we present a supervised learning approach that produces SNNs with high robustness to mismatch and other common sources of noise. Our method trains SNNs to perform temporal classification tasks by mimicking a pre-trained dynamical system, using a local learning rule from non-linear control theory. We demonstrate our method on two tasks requiring temporal memory, and measure the robustness of our approach to several forms of noise and mismatch. We show that our approach is more robust than common alternatives for training SNNs. Our method provides robust deployment of pretrained networks on mixed-signal neuromorphic hardware, without requiring per-device training or calibration. Scientific Reports, 11 (1) ISSN:2045-2322 |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full text from SpringerLink