Toward expert systems on a chip

Abstract

1. INTRODUCTION. In this investigation we attempt to quantify expert reasoning to process boolean values 0 (FALSE) and 1 (TRUE) instead of large lists of atoms that form conditions. The human brain reasons by processing demodulated signals through axons and across synapses, which either fire or don't fire in a boolean fashion. Our approach bridges the current expert systems approach (see, e.g., Hayes-Roth et al.[3]) and the old approaches, e.g., neural nets (McCulloch and Pitts[6]) and associations (Nakano[9]). It simplifies knowledge representation by reducing much of symbolic processing to efficient ANDing and ORing. An ultimate application of our approach would be an expert systems on a chip. A side effect of this would be the demise of inefficient, memory cluttering, list processing languages.

DOI: 10.1145/15923.15927

Extracted Key Phrases

Cite this paper

@article{Looney1986TowardES, title={Toward expert systems on a chip}, author={Carl G. Looney and Abdulrudah R. Alfize}, journal={SIGART Newsletter}, year={1986}, volume={98}, pages={26-28} }