Strategic Concealment of Environment Representations in Competitive Games

2025-09-05

AI Summary

This work investigates how attackers in competitive games strategically conceal their abstract environmental representations to mislead defenders. We propose a novel paradigm—“environmental representation hiding”—formulating the attacker-defender interaction as a Bayesian game and unifying belief inference, strategic planning, and belief manipulation within a bilinear programming framework. Crucially, we introduce an explicit belief manipulation mechanism: the attacker selects controllable trajectories to actively shape the defender’s environmental beliefs, thereby inducing suboptimal obstacle placements. Experiments demonstrate that our approach significantly increases attack success rates, yielding an average strategic advantage improvement of 23.6% across diverse map configurations. These results validate the efficacy and generality of covert cognitive representation hiding as a tractable, optimization-amenable adversarial strategy.

Technology Category

• Game Theory and Economic Paradigms: Adversarial Learning
• Multiagent Systems: Adversarial Agents
• Search and Optimization: Adversarial Search

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

This paper investigates the strategic concealment of map abstractions used by the players in competitive games. We consider a defense scenario in which one player (the Defender) seeks to infer and exploit the abstraction used by the other player (the Attacker). The interaction between the two players is modeled as a Bayesian game: the Defender selects a barrier configuration, i.e., a placement of obstacles that can obstruct the Attacker's movement, based on its belief about the Attacker's abstraction, while the Attacker chooses a trajectory that may intentionally obfuscate its own abstraction of the environment to mislead the Defender. We show that purposeful abstraction concealment naturally emerges from this formulation as a means of improving the Attacker's performance. To solve the game, we propose a bilinear programming approach that integrates Bayesian inference, strategic planning, and belief manipulation. Simulations demonstrate that, by shaping the Defender's belief, the Attacker can induce suboptimal Defender barrier placement, thereby gaining a strategic advantage.

Problem

• Strategic concealment of map abstractions in competitive games
• Defender infers and exploits Attacker's environmental representation
• Attacker obfuscates abstraction to mislead Defender's barrier placement

Innovation

• Bayesian game modeling strategic concealment
• Bilinear programming for Bayesian inference integration
• Attacker manipulates belief for strategic advantage