Proc. Int'l Conf. on Artificial Neural Networks, ICANN'98, Skövde, Sep. 2-4, eds. L. Niklasson, M. Bodén, and T. Ziemke, in series Perspectives in Neural Computing, publ. Springer-Verlag, London, pp. 555-560 (1998-09-02) (bibtex)

Learning invariance manifolds.

Laurenz Wiskott

Abstract: A new algorithm for learning invariance manifolds is introduced that allows a neuron to learn a non-linear transfer function to extract invariant or rather slowly varying features from a vectorial input sequence. This is generalized to a group of neurons, referred to as a Gibson-clique, to learn slowly varying features that are uncorrelated. Since the transfer functions are non-linear, this technique can be applied iteratively to learn more and more complex and invariant features in a hierarchical architecture. Two simple examples demonstrate the general properties of the learning algorithm.

Relevant Project:

May 25, 1998, Laurenz Wiskott,