Conditioned response training of robots using Adaptrode-based neural networks
George E. Mobus, Paul S. Fisher · 1991
Summary form only given, as follows. A simple competitive learning network of Adaptrode-based artificial neurons has been shown to learn a conditioned response behavior. The response characteristics are a function of frequency, duration, phase, and intensity of copresentations of a conditionable stimulus with an unconditionable stimulus. Furthermore, the system is capable of encoding a short-term, contrary association without interference with the long-term association. Mobile, autonomous systems may be trained by conditioned response to perform desired behaviors in remote locations while maintaining an ability to learn from experience.>