Robust Fast Adaptation from Adversarially Explicit Task Distribution Generation

2024-07-28

arXiv.org

AI Summary

In meta-learning, manually designed task distributions often fail to cover critical scenarios, leading to substantial degradation in generalization under task distribution shift. To address this, we propose the first learnable generative task distribution that explicitly models task identities. We further introduce a Stackelberg-game-based adversarial training framework, theoretically guaranteeing robust few-shot adaptation under worst-case distributional shifts. Our approach is compatible with mainstream meta-learners (e.g., MAML, Reptile) and integrates variational inference with generative modeling to enable end-to-end optimization of the task distribution. Extensive experiments on diverse subgroup-shift benchmarks demonstrate consistent and significant improvements over state-of-the-art methods, markedly enhancing cross-distribution generalization. The implementation is publicly available.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Game Theory and Economic Paradigms: Adversarial Learning
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

Meta-learning is a practical learning paradigm to transfer skills across tasks from a few examples. Nevertheless, the existence of task distribution shifts tends to weaken meta-learners' generalization capability, particularly when the training task distribution is naively hand-crafted or based on simple priors that fail to cover critical scenarios sufficiently. Here, we consider explicitly generative modeling task distributions placed over task identifiers and propose robustifying fast adaptation from adversarial training. Our approach, which can be interpreted as a model of a Stackelberg game, not only uncovers the task structure during problem-solving from an explicit generative model but also theoretically increases the adaptation robustness in worst cases. This work has practical implications, particularly in dealing with task distribution shifts in meta-learning, and contributes to theoretical insights in the field. Our method demonstrates its robustness in the presence of task subpopulation shifts and improved performance over SOTA baselines in extensive experiments. The code is available at the project site https://sites.google.com/view/ar-metalearn.

Problem

• Addressing task distribution shifts in meta-learning
• Improving generalization with adversarial task generation
• Enhancing adaptation robustness in worst-case scenarios

Innovation

• Adversarially generated explicit task distributions
• Stackelberg game model for task structure
• Robust fast adaptation in worst cases