How Fine-Tuning Allows for Effective Meta-Learning
arXiv:2105.02221
Abstract
Representation learning has been widely studied in the context of meta-learning, enabling rapid learning of new tasks through shared representations. Recent works such as MAML have explored using fine-tuning-based metrics, which measure the ease by which fine-tuning can achieve good performance, as proxies for obtaining representations. We present a theoretical framework for analyzing representations derived from a MAML-like algorithm, assuming the available tasks use approximately the same underlying representation. We then provide risk bounds on the best predictor found by fine-tuning via gradient descent, demonstrating that the algorithm can provably leverage the shared structure. The upper bound applies to general function classes, which we demonstrate by instantiating the guarantees of our framework in the logistic regression and neural network settings. In contrast, we establish the existence of settings where any algorithm, using a representation trained with no consideration for task-specific fine-tuning, performs as well as a learner with no access to source tasks in the worst case. This separation result underscores the benefit of fine-tuning-based methods, such as MAML, over methods with "frozen representation" objectives in few-shot learning.
References in corpus (28)
- Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
- Prototypical Networks for Few-shot Learning
- Communication-Efficient Learning of Deep Networks from Decentralized Data
- Theoretical Models of Learning to Learn
- Meta-SGD: Learning to Learn Quickly for Few-Shot Learning
- Transfusion: Understanding Transfer Learning for Medical Imaging
- On First-Order Meta-Learning Algorithms
- Meta-Learning with Implicit Gradients
- Continuous Adaptation via Meta-Learning in Nonstationary and Competitive Environments
- Meta-learning with differentiable closed-form solvers
- A PAC-Bayesian bound for Lifelong Learning
- The Benefit of Multitask Representation Learning
- Meta-Learning by Adjusting Priors Based on Extended PAC-Bayes Theory
- How to train your MAML
- On the Theory of Transfer Learning: The Importance of Task Diversity
- Few-Shot Learning via Learning the Representation, Provably
- Provable Meta-Learning of Linear Representations
- Predicting What You Already Know Helps: Provable Self-Supervised Learning
- On the Convergence Theory of Gradient-Based Model-Agnostic Meta-Learning Algorithms
- Convergence of Meta-Learning with Task-Specific Adaptation over Partial Parameters
- Task-Robust Model-Agnostic Meta-Learning
- Incremental Learning-to-Learn with Statistical Guarantees
- Reconciling meta-learning and continual learning with online mixtures of tasks
- Gradients as Features for Deep Representation Learning
- How Important is the Train-Validation Split in Meta-Learning?
- Conditional Meta-Learning of Linear Representations
- How Does the Task Landscape Affect MAML Performance?
- Theoretical bounds on estimation error for meta-learning