This paper introduces a novel approach to multi-agent multi-armed bandit problems within competitive two-sided matching markets. It incorporates Cumulative Prospect Theory (CPT) to model human preferences, which non-linearly weigh agent actions. The research analyzes existing learning algorithms under CPT-distorted rewards, deriving player-optimal regret bounds. Further improvements are proposed to enhance regret guarantees, particularly when the number of arms significantly exceeds the number of players. The study also addresses adversarial market conditions, developing algorithms for robust markets with CPT as a risk-sensitive measure and establishing logarithmic regret guarantees. AI
IMPACT Introduces novel theoretical frameworks for decision-making in competitive AI environments.
RANK_REASON The cluster contains a single academic paper published on arXiv.
- alphaXiv
- behavioral economics
- CatalyzeX Code Finder for Papers
- CORE Recommender
- Cumulative prospect theory
- DagsHub
- Gotit.pub
- Hugging Face
- IArxiv Recommender
- Influence Flower
- multi-agent multi-armed bandit problem
- risk sensitive machine learning
- ScienceCast
- two-sided matching markets
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
