Modular meta-learning
Machine Learning
2019-05-06 v2 Robotics
Machine Learning
Abstract
Many prediction problems, such as those that arise in the context of robotics, have a simplifying underlying structure that, if known, could accelerate learning. In this paper, we present a strategy for learning a set of neural network modules that can be combined in different ways. We train different modular structures on a set of related tasks and generalize to new tasks by composing the learned modules in new ways. By reusing modules to generalize we achieve combinatorial generalization, akin to the "infinite use of finite means" displayed in language. Finally, we show this improves performance in two robotics-related problems.
Cite
@article{arxiv.1806.10166,
title = {Modular meta-learning},
author = {Ferran Alet and Tomás Lozano-Pérez and Leslie P. Kaelbling},
journal= {arXiv preprint arXiv:1806.10166},
year = {2019}
}
Comments
Presented at CoRL 2018