LEARNING TO GENERALIZE FROM SINGLE EXAMPLES IN THE DYNAMIC LINK ARCHITECTURE

A large attraction of neural systems lies in their promise of replacin g programming by learning. A problem with many current neural models i s that with realistically large input patterns learning time explodes. This is a problem inherent in a notion of learning that is based almo st entirely on statistical estimation. We propose here a different lea rning style where significant relations in the input pattern are recog nized and expressed by the unsupervised self-organization of dynamic l inks. The power of this mechanism is due to the very general a priori principle of conservation of topological structure. We demonstrate tha t style with a system that learns to classify mirror symmetric pixel p atterns from single examples.