Stephen Downes

Knowledge, Learning, Community

I remember once writing a post on how neural networks could detect events in time called Working Memory and Object Permanence (more) morivated by a comment from Matthias Melcher. Anyhow, here is a very similar idea expressed in patent form (predating, as far as I can tell, my comment by a few years). Basically, you have a set of neurons that fire at different intervals of time after an event. "This allows, for example, a connection to be specified with a long-term (slow-learning) connection-strength, which will learn directly from a fast-forming connection-strength at the same synapse. The fast-forming connection quickly converges on transient responses, and just as quickly forgets them." Via Reddit.

[Direct link]

files/images/MultiTemporalSynapseNeuron.jpg

Stephen Downes Stephen Downes, Casselman, Canada
stephen@downes.ca

Creative Commons License.

Copyright 2021
Last Updated: Mar 31, 2021 12:00 a.m.