Skip to main content
eScholarship
Open Access Publications from the University of California

UCLA

UCLA Electronic Theses and Dissertations bannerUCLA

Synaptic Resistor Networks for Intelligent Systems with Real-Time Learning

  • Author(s): Tudor, Andrew William
  • Advisor(s): Chen, Yong
  • et al.
Abstract

The human brain is a neural network capable of logically processing massively parallel information while simultaneously improving its performance via learning in dynamic environments. The parallel processing and learning functions are integrated in each synapse, the junction between neurons, but missed by electronic devices. Synaptic resistors, having parallel processing and learning functions, are arranged into a synaptic resistor network to improve an autonomous dynamical system’s performance without any a priori knowledge, and unassisted by additional computing and memory circuits. In this way, the system can exhibit intelligent behavior in a similar way to humans. In comparison with humans learning the same task, the artificially intelligent network exhibited a similar, but slightly superior, strategy, speed, and accuracy, in the simplified, but unknown environment. Synaptic resistor circuits could be scaled up to efficiently process and learn from massively parallel information in dynamic and unknown environments with intelligence comparable to human brains.

Main Content
Current View