Optical implementation of artificial synapses by means of the Wigner distribution approach

A new compact optical processor is proposed for implementing the synaptic interconnections of artificial neural networks. As a consequence of exploiting coordinate transformations, the synaptic elements are compact and applicable to any information. Hence, for use in conjunction with spatial light modulators, the proposed system leads to a programmable neural network. The synaptic interconnections between neurons are performed by means of diffractive optical elements. The required local phase variations of the diffractive optical elements are determined by using the Wigner distribution function. Preliminary experimental verification is presented for the 1-D and the 2-D synapses of neural networks