AUTONOMUS CONTROL STRUCTURE FOR ARTIFICIAL COGNITIVE AGENT

Prof. Roland Hausser
Department of Computational Linguistics
Frederich Alexander University Erlangen, Nueremberg, Germany
Wednesday, February 28th, 15SH, 4:00 pm - 5:00 pm

ABSTRACT:

This paper presents a control structure based on the principle of balance. This principle mediates between the knowledge of the cognitive agent CA and its current situation. CA's knowledge is represented in the form of concatenated propositions. These are individuated into recognition-action-recognition (rac) sequences. Each rac sequence is assigned a need vector indicating whether the associated action raised, lowered, or left unchanged associated physiological or social need parameters. Recognizing a certain combination of concepts, CA activates rac sequences beginning with those concepts, choosing the rac sequence most likely to maintain or regain equilibrium as its model of action in that situation. This general mechanism for keeping the need parameters within normal range may be viewed as a computational implementation of the notion of purpose or intention.

 

  Prof. Roland Hausser studied mathematics and literature in Berlin, and obtained a Ph.D. in theoretical linguistics from the University of Texas at Austin in 1974. He subsequently taught at the University of Munich, where he obtained his habilitation in 1983. He spent a total of 12 years in the US, divided evenly between Austin, Pittsburgh, and Stanford. He is now Professor of Computational Linguistics at the University of Erlangen Nuernberg.