Goal Recognition for Rational and Irrational Agents

Peta Masters
Peta Masters

adaptive agents and multi-agents systems, pp. 440-448, 2019.

Cited by: 10|Bibtex|Views2
EI
Other Links: academic.microsoft.com|dblp.uni-trier.de

Abstract:

Contemporary cost-based goal-recognition assumes rationality: that observed behaviour is more or less optimal. Probabilistic systems, however, generate probability distributions on the basis of suboptimality. We show that, when an observed agent is only slightly irrational (suboptimal), state-of-the-art systems produce counterintuitive re...More

Code:

Data:

Your rating :
0

 

Tags
Comments