Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
Chelsea Finn Pieter Abbeel Sergey Levine
Abstract
We propose an algorithm for meta-learning that is model-agnostic, in the sense that it is compatible with any model trained with gradient descent and applicable to a variety of different learning problems, including classification, regression, and reinforcement learning. The goal of meta-learning is to train a model on a variety of learning tasks, such that it can solve new learning tasks using only a small number of training samples. In our approach, the parameters of the model are explicitly trained such that a small number of gradient steps with a small amount of training data from a new task will produce good generalization performance on that task. In effect, our method trains the model to be easy to fine-tune. We demonstrate that this approach leads to state-of-the-art performance on two few-shot image classification benchmarks, produces good results on few-shot regression, and accelerates fine-tuning for policy gradient reinforcement learning with neural network policies.
中文速览
让神经网络"学会学习"一直是个难题——如何让模型只看几个样本就能迅速掌握新任务？这篇论文提出了 MAML（Model-Agnostic Meta-Learning，与模型无关的元学习）算法：它不是去设计一个专用的学习规则，而是直接优化模型的初始参数，使得模型只需在新任务上做一两步梯度更新，就能获得出色的泛化表现，相当于把模型训练成"极易微调"的状态。这个思路几乎不限制模型结构，全连接网络、卷积网络、循环网络都能用，也适用于分类、回归和强化学习等不同场景。实验表明，MAML 在少样本图像分类基准上达到当时最优水平，在少样本回归上表现良好，在强化学习中也显著加速了策略的适应速度，远超直接预训练的基线。这项工作的重要性在于，它用一个极其简洁统一的框架解决了跨任务快速学习的问题，为后续大量元学习研究奠定了基础。
原文 arXiv:1703.03400;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1703.03400v3