Learning to learn by gradient descent by gradient descent
Marcin Andrychowicz1, Misha Denil1, Sergio Gómez Colmenarejo1, Matthew W. Hoffman1, David Pfau1, Tom Schaul1, Brendan Shillingford1,2, Nando de Freitas1,2,3 1Google DeepMind 2University of Oxford 3Canadian Institute for Advanced Research
Abstract
The move from hand-designed features to learned features in machine learning has been wildly successful. In spite of this, optimization algorithms are still designed by hand. In this paper we show how the design of an optimization algorithm can be cast as a learning problem, allowing the algorithm to learn to exploit structure in the problems of interest in an automatic way. Our learned algorithms, implemented by LSTMs, outperform generic, hand-designed competitors on the tasks for which they are trained, and also generalize well to new tasks with similar structure. We demonstrate this on a number of tasks, including simple convex problems, training neural networks, and styling images with neural art.
中文速览
手工设计优化算法（如SGD、Adam）长期以来是机器学习的主流做法，但这篇论文提出：优化器本身也可以被"学出来"。研究者将设计优化算法的过程转化为一个元学习（meta-learning）问题，用LSTM循环神经网络作为优化器，让它通过大量训练任务自动学会如何更新参数，而无需人工推导更新规则。在二次函数、MNIST神经网络训练、神经风格迁移等多项任务上，这个学出来的优化器不仅明显优于SGD、Adam等手工设计的基线方法，还能泛化到结构相似但未见过的新任务。这项工作的意义在于，它打通了"学习特征"与"学习算法"之间的最后一道壁垒——既然我们已经用学习取代了人工特征工程，现在连优化过程本身也可以从数据中自动习得。
原文 arXiv:1606.04474;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1606.04474v2