Online Bounded Component Analysis: A Simple Recurrent Neural Network with Local Update Rule for Unsupervised Separation of Dependent and Independent Sources

摘要

A low complexity recurrent neural network structure is proposed for unsupervised separation of both independent and dependent sources from their linear mixtures. The proposed network is generated based on Bounded Component Analysis (BCA) approach. We first propose an Online-BCA optimization setting. Then we derive the corresponding recurrent neural network (RNN) with iterative learning update expressions. The resulting 2-layer network has a fairly simple structure with feedforward synapses at the input layer, recurrent synapses at the output layer, and top-down connections from the output layer to the first layer. The synaptic weight updates of the proposed RNN are local, supporting its biological plausibility. We use correlated synthetic sources and natural images as examples to illustrate the correlated/dependent source separation capability of the proposed neural network.