GST networks: learning emergent spatiotemporal correlations
Chaitanya Tumuluri, Chilukuri K. Mohan, A.N. Choudhary · 2002
This paper presents two novel networks-the growing cell structure instantaneous spatio-temporal (GIST) network, and growing cell structure epochal spatio-temporal (GEST) network-which combine unsupervised feature-extraction and Hebbian learning, for tracking emergent correlations in the evolution of spatio-temporal distributions. The networks were successfully tested on the challenging data mapping problem, using an execution driven simulation of their implementation in hardware.