RECONFIGURABLE AND CUSTOMIZABLE GENERAL-PURPOSE CIRCUITS FOR NEURAL NETWORKS

摘要

A reconfigurable neural network circuit is provided. The reconfigurable neural network circuit comprises an electronic synapse array including multiple synapses interconnecting a plurality of digital electronic neurons. Each neuron comprises an integrator that integrates input spikes and generates a signal when the integrated inputs exceed a threshold. The circuit further comprises a control module for reconfiguring the synapse array. The control module comprises a global final state machine that controls timing for operation of the circuit, and a priority encoder that allows spiking neurons to sequentially access the synapse array.

主权项

1. A method for non-linear pattern classification of images, comprising:
in a learning phase:
receiving an input image; andbased on the input image, training a neural network to learn and classify a pattern included in the input image, wherein the neural network comprises a plurality of neurons interconnected via a plurality of synapses; and in a recall phase:
receiving corrupted data; andperforming pattern recognition on the corrupted data, wherein the pattern is recalled if the corrupted data comprises an incomplete version of the pattern.