Title: Dynamic Links as Basis for Brain Architecture

Christoph von der Malsburg

Computer Science Department and Program in Neuroscience

University of Southern California, Los Angeles and

Institut für Neuroinformatik, Ruhr-Universität Bochum

Abstract:

To understand the brain, four fundamental questions must be asked first:

a) what is the data structure of brain states? b) how are brain states organized? c) what is the data structure of memory? and d) what is the mechanism of learning, that is, how is memory organized? Artificial neural networks, from McCulloch & Pitts to backprop and beyond, give definite answers to those questions. I would like to take issue with this widely accepted view (except regarding question c) and discuss the Dynamic Link Architecture as alternative. I will illustrate the power of the DLA by applications to invariant object recognition. A very close relationship to shifter circuits will become evident.