Rectified Gaussian distributions and the identification of multiple cause structure in data

摘要

We investigate the use of an unsupervised artificial neural network to form a sparse representation of the underlying causes in a data set. By using fixed lateral connections that are derived from the Rectified Generalised Gaussian distribution,we form a network that is capable of identifying the multiple cause structure of the data. We further show that some topology preservation of the input data is possible using this network and that related features may be coded in separate areas of theoutput space.