Coevolvable hardware platform for automatic hardware design of neural networks

Abstract

This paper proposes a system for the automatic hardware design of neural networks based on cooperative coevolutionary paradigms and multiple reconfigurable devices. The authors' system is composed of a logic synthesis tool, multiple reconfigurable devices and an embedded processor executing the coevolutionary algorithm. The partitioning of the engineering design process follows current practices in hardware/software codesign based on both information on arrival rate /spl lambda/ of requests and the service time /spl mu/ of the reconfigurable devices. The system is suitable under some conditions for industrial applications such as a reactive system but also because it can be connected to multiple systems in a totally networked industrial environment which allows download of the same hardware configuration on multiple on ine devices.