Improving modeling of other agents using tentative stereotypes and compactification of observations

Abstract

We investigate possible improvements to modeling other agents based on observed situation-action pairs and the nearest neighbor rule. Tentative stereotype models allow for good predictions of a modeled agent's behavior even after few observations. Periodic revaluation of the chosen stereotype and the potential for switching between different stereotypes or to the observation based model aids in dealing with very similar (but not identical) stereotypes and agents that do not conform to any stereotype. Finally, compactification of observations keeps the application of the model efficient by reducing comparisons within the nearest neighbor rule. Our experiments show that stereotyping significantly improves cases where using just the original method performs badly and that revaluation and switching fortify stereotyping against the potential risk of using an incorrect stereotype. Compactification shows good potential for improving efficiency, but is sometimes at risk of losing important observations.