Independent Reading · 2022
Inverse Game Theory for Stackelberg Games: The Blessing of Bounded Rationality
Jibang Wu, Weiran Shen, Fei Fang, and Haifeng Xu · NeurIPS
Stackelberg GamesInverse Game TheoryBounded Rationality
Full text
Read the Paper
Use View PDF or Download PDF to read the full paper on this device.
Paper Information
- Authors
- Jibang Wu, Weiran Shen, Fei Fang, and Haifeng Xu
- Venue
- NeurIPS · 2022
Summary
This paper studies how a leader can infer unknown follower utilities from observed behavior in Stackelberg games.
Core Idea
Quantal-response behavior can reveal more information than perfectly rational best responses because every action may be observed with positive probability.
Method
The authors derive learning guarantees under quantal response and support the theory with experiments on synthetic games and model misspecification.
Key Takeaways
- Bounded rationality can improve identifiability in inverse strategic problems.
- Smooth behavioral responses provide richer information about latent utilities.
- Behavioral assumptions affect both prediction quality and sample efficiency.
Questions
- How can the rationality parameter be estimated jointly with follower utilities?
- Do the benefits persist under non-logit behavioral models?