Cargando…
A solution to the learning dilemma for recurrent networks of spiking neurons
Recurrently connected networks of spiking neurons underlie the astounding information processing capabilities of the brain. Yet in spite of extensive research, how they can learn through synaptic plasticity to carry out complex network computations remains unclear. We argue that two pieces of this p...
Autores principales: | , , , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Nature Publishing Group UK
2020
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7367848/ https://www.ncbi.nlm.nih.gov/pubmed/32681001 http://dx.doi.org/10.1038/s41467-020-17236-y |
_version_ | 1783560496659562496 |
---|---|
author | Bellec, Guillaume Scherr, Franz Subramoney, Anand Hajek, Elias Salaj, Darjan Legenstein, Robert Maass, Wolfgang |
author_facet | Bellec, Guillaume Scherr, Franz Subramoney, Anand Hajek, Elias Salaj, Darjan Legenstein, Robert Maass, Wolfgang |
author_sort | Bellec, Guillaume |
collection | PubMed |
description | Recurrently connected networks of spiking neurons underlie the astounding information processing capabilities of the brain. Yet in spite of extensive research, how they can learn through synaptic plasticity to carry out complex network computations remains unclear. We argue that two pieces of this puzzle were provided by experimental data from neuroscience. A mathematical result tells us how these pieces need to be combined to enable biologically plausible online network learning through gradient descent, in particular deep reinforcement learning. This learning method–called e-prop–approaches the performance of backpropagation through time (BPTT), the best-known method for training recurrent neural networks in machine learning. In addition, it suggests a method for powerful on-chip learning in energy-efficient spike-based hardware for artificial intelligence. |
format | Online Article Text |
id | pubmed-7367848 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2020 |
publisher | Nature Publishing Group UK |
record_format | MEDLINE/PubMed |
spelling | pubmed-73678482020-07-21 A solution to the learning dilemma for recurrent networks of spiking neurons Bellec, Guillaume Scherr, Franz Subramoney, Anand Hajek, Elias Salaj, Darjan Legenstein, Robert Maass, Wolfgang Nat Commun Article Recurrently connected networks of spiking neurons underlie the astounding information processing capabilities of the brain. Yet in spite of extensive research, how they can learn through synaptic plasticity to carry out complex network computations remains unclear. We argue that two pieces of this puzzle were provided by experimental data from neuroscience. A mathematical result tells us how these pieces need to be combined to enable biologically plausible online network learning through gradient descent, in particular deep reinforcement learning. This learning method–called e-prop–approaches the performance of backpropagation through time (BPTT), the best-known method for training recurrent neural networks in machine learning. In addition, it suggests a method for powerful on-chip learning in energy-efficient spike-based hardware for artificial intelligence. Nature Publishing Group UK 2020-07-17 /pmc/articles/PMC7367848/ /pubmed/32681001 http://dx.doi.org/10.1038/s41467-020-17236-y Text en © The Author(s) 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. |
spellingShingle | Article Bellec, Guillaume Scherr, Franz Subramoney, Anand Hajek, Elias Salaj, Darjan Legenstein, Robert Maass, Wolfgang A solution to the learning dilemma for recurrent networks of spiking neurons |
title | A solution to the learning dilemma for recurrent networks of spiking neurons |
title_full | A solution to the learning dilemma for recurrent networks of spiking neurons |
title_fullStr | A solution to the learning dilemma for recurrent networks of spiking neurons |
title_full_unstemmed | A solution to the learning dilemma for recurrent networks of spiking neurons |
title_short | A solution to the learning dilemma for recurrent networks of spiking neurons |
title_sort | solution to the learning dilemma for recurrent networks of spiking neurons |
topic | Article |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7367848/ https://www.ncbi.nlm.nih.gov/pubmed/32681001 http://dx.doi.org/10.1038/s41467-020-17236-y |
work_keys_str_mv | AT bellecguillaume asolutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT scherrfranz asolutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT subramoneyanand asolutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT hajekelias asolutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT salajdarjan asolutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT legensteinrobert asolutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT maasswolfgang asolutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT bellecguillaume solutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT scherrfranz solutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT subramoneyanand solutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT hajekelias solutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT salajdarjan solutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT legensteinrobert solutiontothelearningdilemmaforrecurrentnetworksofspikingneurons AT maasswolfgang solutiontothelearningdilemmaforrecurrentnetworksofspikingneurons |