Initial Work on the Development of a Hardware-Based Gradient Descent Trained Expert System

Prior work has introduced a form of explainable artificial intelligence that is able to precisely explain, in human-understandable form, why it makes decisions. It also learn make better decisions without potentially learning illegal or invalid considerations. This defensible system based on fractional value rule-fact expert systems and the use gradient descent training optimize rule weightings. software demonstrated efficacy for many applications; however, utilizes iterative processing thus does not have deterministic completion time. requires comparatively expensive general-purpose computing hardware run on. paper builds prior development hardware-based presents assesses implementation this system. characterizes its performance discusses utility trade-offs several application domains.