Backpropagation through space, time, and the brain

arXiv:2403.16933v1 Announce Type: cross
Abstract: Effective learning in neuronal networks requires the adaptation of individual synapses given their relative contribution to solving a task. However, physical neuronal systems -- whether biological or artificial -- are constrained by spatio-temporal locality. How such networks can perform efficient credit assignment, remains, to a large extent, an open question. In Machine Learning, the answer is almost universally given by the error backpropagation algorithm, through both space (BP) and time (BPTT). However, BP(TT) is well-known …

abstract artificial arxiv backpropagation brain credit cs.ai cs.lg cs.ne eess.sp however networks q-bio.nc space synapses systems temporal through type