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 advantage of this method is that the clipping point is adaptable to the stored palter ns. By introducing an orthogonal projection algorithm in the Hopfield model and using the adaptive clipping method. an adaptive clipped model is constructed. Computer simulations for the reconstruction of similar traffic signs embedded in the random noise by the adaptive clipped model have shown improved performance compared with the Hopfield model and the former clipped models. A 2-D hybrid optical neural network with 1024 neurons is proposed to implement the adaptive clipped model, in which a beam-direction encoding-method is employed, Preliminary experimental results for the reconstruction of traffic signs are shown.