An integrated circuit for artificial neural networks
A brief introduction is made of the Backpropagation algorithm and its parallelization in a proposed architecture. The main requirements of each of the architecture's processors is then elaborated and the final Integrated Circuit design presented. Each part of the processor is discussed separately: ALU, communications unit, memory and the control unit. The control unit is discussed with more detail by explaining how it performs when doing the emulation.
Unable to display preview. Download preview PDF.
- F.Castillo, J.Cabestany, "A VLSI Neural Net Architecture — A Proposal", Proc. Neuro-Nîmes'90.Google Scholar
- F.Castillo, J.M.Moreno, J.Cabestany, "Digital VLSI Implementation of a Neural Processor", Proc. Melecon '91, Yugoslavia.Google Scholar
- F.Castillo, P.Amengual, J.Cabestany, "A Sensibility Study of the Backpropagation Algorithm", Proc. ICANN'91, Helsinki.Google Scholar