Execution Skill Estimation

In domains with continuous action spaces, one characteristic of an agent is their precision in executing intended actions. An agent's execution skill significantly impacts their success as it determines how much executed actions deviate from intended actions. We introduce the problem of estimating an agent's execution skill level given only observations of their executed actions. The main difficulty is that while executed actions are observed, the intended actions are not, thus the amount of action deviation due to imperfect execution skill is not obvious. We introduce a simple experimental domain in which this problem can be studied and present a method that focuses on observed rewards to estimate execution skill. This method is experimentally evaluated and shown be able to estimate an agent's execution skill under certain conditions.

[1]  Christopher Archibald,et al.  A Distributed Agent for Computational Pool , 2016, IEEE Transactions on Computational Intelligence and AI in Games.

[2]  Michael H. Bowling,et al.  Action Selection for Hammer Shots in Curling , 2016, IJCAI.

[3]  Pieter Spronck,et al.  Opponent Modeling in Real-Time Strategy Games , 2007, GAMEON.

[4]  Manuela M. Veloso,et al.  Existence of Multiagent Equilibria with Limited Agents , 2004, J. Artif. Intell. Res..

[5]  David Sarne,et al.  Agent-Mediated Electronic Commerce. Designing Trading Strategies and Mechanisms for Electronic Markets , 2014, Lecture Notes in Business Information Processing.

[6]  David Carmel,et al.  Opponent Modeling in Multi-Agent Systems , 1995, Adaption and Learning in Multi-Agent Systems.

[7]  Michael H. Bowling,et al.  Using Response Functions to Measure Strategy Strength , 2014, AAAI.

[8]  Sarvapali D. Ramchurn,et al.  Continuous Double Auctions with Execution Uncertainty , 2009, AMEC/TADA.

[9]  Christopher Archibald,et al.  Success, strategy and skill: an experimental study , 2010, AAMAS.

[10]  Michael H. Bowling,et al.  Online implicit agent modelling , 2013, AAMAS.

[11]  Christopher Archibald,et al.  Computational Pool: A New Challenge for Game Theory Pragmatics , 2010, AI Mag..

[12]  Manish Jain,et al.  Risk-Averse Strategies for Security Games with Execution and Observational Uncertainty , 2011, AAAI.

[13]  Michael H. Bowling,et al.  Decision-Theoretic Clustering of Strategies , 2015, AAAI Workshop: Computer Poker and Imperfect Information.

[14]  Sarit Kraus,et al.  Game-theoretic randomization for security patrolling with dynamic execution uncertainty , 2013, AAMAS.

[15]  Jonathan Schaeffer,et al.  Opponent Modeling in Poker , 1998, AAAI/IAAI.

[16]  Christopher Archibald,et al.  Hustling in Repeated Zero-Sum Games with Imperfect Execution , 2011, IJCAI.

[17]  Nathan R. Sturtevant,et al.  Prob-Maxn: Playing N-Player Games with Opponent Models , 2006, AAAI.

[18]  Jean-Pierre Dussault,et al.  A Straight Approach to Planning for 14.1 Billiards , 2016, IEEE Transactions on Computational Intelligence and AI in Games.

[19]  Michael H. Bowling,et al.  Monte Carlo Tree Search in Continuous Action Spaces with Execution Uncertainty , 2016, IJCAI.