7. Learning in Neural Networks

By Petrut Bogdan, University of Manchester, UK | Garibaldi Pineda García, University of Sussex, UK | Michael Hopkins, University of Manchester, UK | Edward Jones, University of Manchester, UK | James Knight, University of Sussex, UK | Adam Perrett, University of Manchester, UK

Published: 31 Mar 2020

© 2020 Petrut Bogdan | Garibaldi Pineda García | Michael Hopkins | Edward Jones | James Knight | Adam Perrett

Abstract

This chapter is concerned with the motivation, design and implementation behind mimicking biological learning rules with a focus on, you guessed it, SpiNNaker. It starts by presenting Spike-timing-dependent plasticity (STDP) operating in an unsupervised fashion based on relative spike times of the pre- and post-synaptic neurons or based on the sub-threshold membrane potential. This is followed by a model of STDP modulated by the presence of an additional signal and operating on eligibility traces. Longer-term mechanisms in the form of structural plasticity, involving the rewiring of connections between the neurons, and (very long-term) neuroevolution close out the chapter.