HARDWARE REQUIREMENTS FOR NEURAL NETWORK PATTERN CLASSIFIERS - A CASE-STUDY AND IMPLEMENTATION

A special-purpose chip, optimized for computational needs of neural networks, performs over 2,000 multiplications and additions simultaneously. Its data path is suitable particularly for the convolutional architectures typical in pattern classification networks but can also be configured for fully connected or feedback topologies. A development system permits rapid prototyping of new applications and analysis of the impact of the specialized hardware on system performance. We demonstrate the power and flexibility of the processor with a neural network for handwritten character recognition containing over 133,000 connections.