On Learning Action Costs from Input Plans
Published in Proceedings of the European Conference on Artificial Intelligence (ECAI 2025), 2025
Recommended citation: Morales, M., Pozanco, A., Canonaco, G., Gopalakrishnan, S., Borrajo, D., & Veloso, M. (2025). On learning action costs from input plans. In Proceedings of the European Conference on Artificial Intelligence (ECAI).
Abstract
Understanding user preferences and implicit cost trade-offs in automated planning often requires inferring underlying cost functions directly from observed behavior. We formulate the problem of learning numeric action costs from plan traces using mathematical programming and linear constraint satisfaction, demonstrating rapid convergence to expert cost structures.
Recommended citation: Morales, M., Pozanco, A., Canonaco, G., Gopalakrishnan, S., Borrajo, D., & Veloso, M. (2025). On learning action costs from input plans. In Proceedings of the European Conference on Artificial Intelligence (ECAI).