GRADED LEARNING DEVICE AND METHOD

A graded-learning processing network which grades its performance as it maps the input-output relationship during a training period. At the end of an operation of the Processing Newtork, its performance is graded and network variables are adjusted or amended, and the processing network is operated again and its performance is graded. The order or direction of performance grades (e.g. better or worse) are noted and the adjustments or amendments of the processing network may proceed in the same direction (or opposite direction) depending upon the grading of subsequent performances. This obviates the need for information about a desired response on output performance of the processing network at any given time, and is conducive to efficient learning for improved performance in a processing environment in which the operational parameters are not known.