Self-Organization and Association for Fine Spatio-Temporal Spike Sequences

Abstract

In this paper, we discuss unsupervised learning for a temporally precise sequence. A network of leaky integrate-and-fire neurons is able to learn a fine spatio-temporal pattern, when the neurons are provided many excitatory random inputs. This unsupervised learning is achieved by selecting appropriate connections in the network. After learning, the trained network works as an associative memory with high temporal precision. Namely, it distinguishes the training sequence through filtering the disarranged sequence according to its correlation value with the training sequence.