Neural Networks and Physical Systems With Emergent Collective Computational Abilities

Exploring Emergent Collective Properties in Neural Networks and Physical Systems

Abstract

Computational properties of use to biological organismsor to the construction of computers can emerge as collective properties ofsystems -having a large number of simple equivalent components (or neurons).The physical meaning of content-addressable memory is described by anappropriate phase space flow of the state of a system. A model of such a systemis given, based on aspects of neurobiology but readily adapted to integratedcircuits. The collective properties of this model produce a content-addressablememory which correctly yields an entire memory from any subpart of sufficientsize. The algorithm for the time evolution of the state of the system is basedon asynchronous parallel processing. Additional emergent collective propertiesinclude some capacity for generalization, familiarity recognition,categorization, error correction, and time sequence retention. The collectiveproperties are only weakly sensitive to details of the modeling or the failureof individual devices.