Topological map from a photorefractive self-organizing neural network

We propose and construct an opto-electronic self-organizing neural network able to create a topological map from a learning set of vectors. After the iterative presentation of all vectors, the input of one vector activates a few neurons of the 2D output whose positions indicate the correlation of this vector with the other vectors of the learning set. The system benefits from the specific kinetics of the photorefractive effect to holographically record and update the neural interconnection weights in a photorefractive crystal.