Learning to learn by gradient descent by gradient descent

• DOI: 10.48550/arxiv.1606.04474
• Publication Date: 2016-01-01
• Authors: Andrychowicz, Marcin; Denil, Misha; Gomez, Sergio; Hoffman, Matthew W.; Pfau, David; Schaul, Tom; Shillingford, Bre...; de Freitas, Nando
• Keywords: FOS: Computer and information sciences; Computer Science - Machine Learning; 0202 electrical engineering, electronic engineering, information engineering; Computer Science - Neural and Evolutionary Computing; 02 engineering and technology; Neural and Evolutionary Computing (cs.NE); Machine Learning (cs.LG)

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.