Curiosity Meets Cooperation: A Game-Theoretic Approach to Long-Tail Multi-Label Learning
It addresses the challenge of improving performance on rare labels in multi-label prediction, which is crucial for practical applications, though it appears incremental as it builds on existing game-theoretic and curiosity-driven concepts.
The paper tackles the problem of long-tail imbalance in multi-label learning, where rare labels are often ignored, by proposing a cooperative game-theoretic framework that achieves state-of-the-art gains, including up to +4.3% Rare-F1 and +1.6% P@3 over baselines.
Long-tail imbalance is endemic to multi-label learning: a few head labels dominate the gradient signal, while the many rare labels that matter in practice are silently ignored. We tackle this problem by casting the task as a cooperative potential game. In our Curiosity-Driven Game-Theoretic Multi-Label Learning (CD-GTMLL) framework, the label space is split among several cooperating players that share a global accuracy payoff yet earn additional curiosity rewards that rise with label rarity and inter-player disagreement. These curiosity bonuses inject gradient on under-represented tags without hand-tuned class weights. We prove that gradient best-response updates ascend a differentiable potential and converge to tail-aware stationary points that tighten a lower bound on the expected Rare-F1. Extensive experiments on conventional benchmarks and three extreme-scale datasets show consistent state-of-the-art gains, delivering up to +4.3% Rare-F1 and +1.6% P@3 over the strongest baselines, while ablations reveal emergent division of labour and faster consensus on rare classes. CD-GTMLL thus offers a principled, scalable route to long-tail robustness in multi-label prediction.