Conventional, stored program architecture systems are designed for algorithmic
and exact calculations. However, the problems with highest impact involve
large, noisy, incomplete, “natural” data sets that do not lend themselves to
convenient solutions by current systems. Our task is to build upon the
convergence among neuroscience, microelectronics and computational systems to
develop a new architecture designed to handle these natural data sets. The
applications and clarification of the value proposition for new neuro-inspired,
neuromorphic systems are critical focal points of this workshop.
Neuroscience: Sensory information processing in cells and circuits,
mechanisms of plasticity, learning and development.
Theory: Theoretical principles of brain information processing, sparse
coding, stochastic computing, the role of spikes, Bayesian computing.
Algorithms: Computational synthesis of brain information processing, deep
Platforms/Hardware: Massively parallel neuromorphic hardware architectures,
application of commodity systems, novel digital, analog and mixed-signal
architectures, application of novel devices.
Applications: Robotics, spatio-temporal pattern detection, causal relations
in big data, prediction, approximate computing.