Independent Reading · 2019
Imitative Follower Deception in Stackelberg Games
Jiarui Gan, Haifeng Xu, Qingyu Guo, Long Tran-Thanh, Zinovi Rabinovich, and Michael Wooldridge · ACM EC
Stackelberg GamesDeceptionAlgorithmic Game Theory
Paper Information
- Authors
- Jiarui Gan, Haifeng Xu, Qingyu Guo, Long Tran-Thanh, Zinovi Rabinovich, and Michael Wooldridge
- Venue
- ACM EC · 2019
Summary
This paper examines how a strategic follower can imitate another type and manipulate a leader that is learning from observed responses.
Core Idea
Follower behaviour should not automatically be treated as truthful evidence about underlying preferences or payoffs.
Method
The authors formalise imitative follower deception, analyse the computational complexity of optimal leader policies, and evaluate deception-aware strategies in simulation.
Key Takeaways
- Information asymmetry can create incentives for strategic misrepresentation.
- Learning a follower model is harder when observations are deliberately manipulated.
- Deception-aware policies can reduce the leader’s loss, but may be computationally difficult to obtain.
Questions
- Which observable signals help distinguish genuine behaviour from imitation?
- How can deception-aware policies remain effective when the follower type space is incomplete?