Feedforward ANN for 2-1 fixed point ALUs
Simon Vassiliadis, Koen Bertels, G.G. Pechanek · 2002
Investigates the possibility of constructing fixed point units using feedforward neural networks. The authors investigate the possibility of constructing small depth neural networks for operations usually defined in general purpose computer architectures. In particular the authors show that fixed operations require no more depth than the networks for binary addition. The authors show that depth-3 networks with bounded weights and small size requirements can be constructed that guarantee architectural compliance for fixed point arithmetic and address generation operations. Consequently, it is suggested that the proposed scheme can be used to potentially produce high performance arithmetic devices for fixed point processing units with small size requirements.