A solution to the learning dilemma for recurrent networks of spiking neurons

Read:: - [ ] Bellec et al. (2020) - A solution to the learning dilemma for recurrent networks of spiking neurons ➕2024-10-14 !!2 rd citation todoist Print::  ❌ Zotero Link:: Zotero Files:: attachment Reading Note:: Web Rip:: url:: https://www.nature.com/articles/s41467-020-17236-y

TABLE without id
file.link as "Related Files",
title as "Title",
type as "type"
FROM "" AND -"Obsidian Assets"
WHERE citekey = "bellecSolutionLearningDilemma2020" 
SORT file.cday DESC

Abstract

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.

Quick Reference

Top Notes

Tasks