Brian Box

About

Brian Box is a pioneering figure in the field of neural network hardware implementation, with a particular focus on the Cerebellar Model Articulation Controller (CMAC). His most notable contribution, the 1990 paper "Design and Implementation of a High Speed CMAC Neural Network Using Programmable CMOS Logic Cell Arrays," introduced a groundbreaking approach to accelerating neural computation through dedicated CMOS logic. Box designed and built two general-purpose CMAC associative memory boards for the VME bus, each capable of supporting up to eight independent CMAC networks with a total of one million adjustable weights. This work demonstrated the feasibility of high-speed, hardware-based neural processing at a time when software implementations were the norm. Although his citation count of 11 reflects the niche nature of early hardware neural networks, Box's contributions were foundational for real-time control and robotics applications, where CMAC's fast learning and generalization capabilities proved invaluable. His research bridged the gap between theoretical neural models and practical, high-performance systems, inspiring subsequent work in embedded AI and neuromorphic engineering.

Research Focus

Key Achievements

Top Papers

1. 1
   Design and Implementation of a High Speed CMAC Neural Network Using Programmable CMOS Logic Cell Arrays

   11 citations · 1990

Key Collaborators

Contact & Links