ADAPTIVE CLIPPED MODEL AND ITS OPTICAL IMPLEMENTATION

An adaptive clipping algorithm for constructing a bipolar and binary ( +1, 0. -1) interconnection weight matrix (IWM) is proposed. The advant age of this method is that the clipping point is adaptable to the stor ed palter ns. By introducing an orthogonal projection algorithm in the Hopfield model and using the adaptive clipping method. an adaptive cl ipped model is constructed. Computer simulations for the reconstructio n of similar traffic signs embedded in the random noise by the adaptiv e clipped model have shown improved performance compared with the Hopf ield model and the former clipped models. A 2-D hybrid optical neural network with 1024 neurons is proposed to implement the adaptive clippe d model, in which a beam-direction encoding-method is employed, Prelim inary experimental results for the reconstruction of traffic signs are shown.