Minimax Game

Minimax games, a framework where two players with opposing objectives compete, are central to many machine learning problems. Current research focuses on improving the training stability and convergence of these games, particularly in adversarial training and generative adversarial networks (GANs), often employing novel algorithms like recursive reasoning methods to approximate optimal strategies. This research addresses challenges like robust overfitting and the need for finding mixed equilibria, impacting the development of robust and fair machine learning models across diverse applications, including image generation and knowledge graph completion. The development of efficient and provably convergent algorithms for solving minimax games is a key area of ongoing investigation.

Papers

October 30, 2023

Balance, Imbalance, and Rebalance: Understanding Robust Overfitting from a Minimax Game Perspective
Yifei Wang, Liangchen Li, Jiansheng Yang, Zhouchen Lin, Yisen Wang
Adversarial Training Balancing Strategy Group Imbalance Robust Overfitting Minimax Game

October 29, 2022

Recursive Reasoning in Minimax Games: A Level $k$ Gradient Play Method
Zichu Liu, Lacra Pavel
Generative Adversarial Network Practical Method Gradient Norm GAN Training Different Level Convergence Behavior Recursive Thinking Minimax Game

June 30, 2022

A note on large deviations for interacting particle dynamics for finding mixed equilibria in zero-sum games
Viktor Nilsson, Pierre Nyquist
Short Note Zero Sum Game Particle Dynamic Large Deviation Equilibrium Bandit Minimax Game

May 2, 2022

Positive-Unlabeled Learning with Adversarial Data Augmentation for Knowledge Graph Completion
Zhenwei Tang, Shichao Pei, Zhao Zhang, Yongchun Zhu, Fuzhen Zhuang, Robert Hoehndorf, Xiangliang Zhang
LeArning Abstract Knowledge Graph Completion Positive Unlabeled Learning Adversarial Data Augmentation Negative Learning Minimax Game

January 28, 2022

Linear Adversarial Concept Erasure
Shauli Ravfogel, Michael Twiton, Yoav Goldberg, Ryan Cotterell
Pre Trained Representation Convex Relaxation Minimax Game

December 12, 2021

Fairness for Robust Learning to Rank
Omid Memarrast, Ashkan Rezaei, Rizal Fathony, Brian Ziebart
Procedural Fairness Group Fairness Robust Learning Stable Rank Fairness Constraint Distributional Robustness Proportional Ranking Minimax Game