Model-Agnostic Learning to Meta-Learn

Abstract

In this paper, we propose a learning algorithm that enables a model to quickly exploit commonalities among related tasks from an unseen task distribution, before quickly adapting to specific tasks from that same distribution. We investigate how learning with different task distributions can first improve adaptability by meta-finetuning on related tasks before improving goal task generalization with finetuning. Synthetic regression experiments validate the intuition that learning to meta-learn improves adaptability and consecutively generalization. Experiments on more complex image classification, continual regression, and reinforcement learning tasks demonstrate that learning to meta-learn generally improves task-specific adaptation. The methodology, setup, and hypotheses in this proposal were positively evaluated by peer review before conclusive experiments were carried out.

arXiv commentPublished in Proceedings of Machine Learning Research, PMLR 148:155-175

References in corpus (14)

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Trust Region Policy Optimization
• Meta-SGD: Learning to Learn Quickly for Few-Shot Learning
• On First-Order Meta-Learning Algorithms
• Meta-Learning with Implicit Gradients
• A Closer Look at Few-shot Classification
• Hierarchically Structured Meta-learning
• Multimodal Model-Agnostic Meta-Learning via Task-Aware Modulation
• Unsupervised Meta-Learning For Few-Shot Image Classification
• BOIL: Towards Representation Change for Few-shot Learning
• Automated Relational Meta-learning
• Toward Multimodal Model-Agnostic Meta-Learning
• Continual Reinforcement Learning with Multi-Timescale Replay