Emerging neuromorphic computing architectures & enabling hardware for cognitive information processing applications

Abstract

The highly cross-disciplinary emerging field of neuromorphic computing architectures for cognitive information processing applications requires knowledge within many research fields: computer architecture, neuroscience, cognitive psychology, cognitive modeling, dynamical systems, belief systems, software, computer engineering, etc. In our effort to develop cognitive systems atop a neuromorphic computing architecture, we explored the issues associated with mapping computing strategies such as the Brain State-in-a-Box and Confabulation within a Cell-BE powered 54 TeraFlops high performance computer Linux cluster. In this work, we seek to understand the underlying mechanisms for emulating neuromorphic-based cognitive process and their computational scaling properties towards human-like cognition and perception.