Neural network with canonical pulsed neurons for a space-time associative memory
暂无分享,去创建一个
Embodiments of the invention relate to canonical pulsed neurons for a spatio-temporal associative memory. One aspect of the invention provides a space-time associative memory ready, which includes a plurality of electronic neurons with a layered relationship in a neural network with directional synaptic connectivity. The plurality of electronic neurons are configured so as to detect the occurrence of a spatio-temporal pattern in real time in a data stream and filter out the spatiotemporal pattern. The plurality of electronic neurons are further configured so as to provide on the basis of learning rules, the spatio-temporal patterns in the plurality of electronic neurons and retrieve the stored spatiotemporal pattern when them a version of the spatio-temporal pattern is presented.
[1] Konstantin K. Likharev,et al. Hybrid CMOS/Nanoelectronic Circuits: Opportunities and Challenges , 2008 .
[2] L. Abbott,et al. Competitive Hebbian learning through spike-timing-dependent synaptic plasticity , 2000, Nature Neuroscience.