Model-Agnostic Learning to Meta-Learn
This work addresses the problem of improving model adaptability and generalization for various machine learning tasks, which is an incremental improvement for researchers and practitioners working with meta-learning.
This paper proposes a learning algorithm that allows a model to quickly adapt to new tasks by first exploiting commonalities among related tasks from an unseen distribution and then fine-tuning on specific tasks. Synthetic regression, image classification, continual regression, and reinforcement learning experiments demonstrate improved task-specific adaptation.
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.