On First-Order Meta-Learning Algorithms
Alex Nichol and Joshua Achiam and John Schulman OpenAI {alex, jachiam,
Abstract
This paper considers meta-learning problems, where there is a distribution of tasks, and we would like to obtain an agent that performs well (i.e., learns quickly) when presented with a previously unseen task sampled from this distribution. We analyze a family of algorithms for learning a parameter initialization that can be fine-tuned quickly on a new task, using only first-order derivatives for the meta-learning updates. This family includes and generalizes first-order MAML, an approximation to MAML obtained by ignoring second-order derivatives. It also includes Reptile, a new algorithm that we introduce here, which works by repeatedly sampling a task, training on it, and moving the initialization towards the trained weights on that task. We expand on the results from Finn et al. showing that first-order meta-learning algorithms perform well on some well-established benchmarks for few-shot classification, and we provide theoretical analysis aimed at understanding why these algorithms work.
中文速览
元学习（meta-learning）领域长期依赖需要计算高阶导数的MAML算法，计算成本高且实现复杂。这篇文章专注于只用一阶梯度的元学习方法，系统分析了一阶MAML（FOMAML），并提出了一个新算法Reptile——它的做法极其简单：反复采样一个任务、在上面训练几步，然后把初始化参数朝训练后的权重方向移动。理论分析表明，FOMAML和Reptile的更新方向都包含两类项：一类等价于联合训练（最小化所有任务的平均损失），另一类更关键，专门最大化同一任务不同数据批次之间梯度的内积，从而促进任务内的泛化能力，这正是"学会快速学习"的核心机制。在Mini-ImageNet和Omniglot少样本分类基准上，两种一阶方法均取得与完整MAML相当的性能，而Reptile连任务内的训练集/测试集划分都不需要，实现更简洁，对那些高阶梯度计算代价过高的场景具有重要的实用价值。
原文 arXiv:1803.02999;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1803.02999v3