Task-Linear Deep Representation of Physical Systems

Matthieu Blanke, Marc Lelarge

Keywords: meta-learning, multi-task learning, maml, interpretable learning, physical systems, robotics

TL;DR: We propose a meta-learning architecture leveraging the structure of physical systems and enabling interpretability of the learned parameters.

Abstract: Machine learning methods can be a valuable aid in the scientific process, but they need to face challenging settings where data come from inhomogeneous experimental conditions. Recent meta-learning methods have made significant progress in multi-task learning, but they rely on black-box neural networks and suffer from a lack of interpretability. We introduce Task-Linear Deep Representation, or TDLR, a new meta-learning architecture capable of learning efficiently from multiple environments by incorporating the linear structure observed in many problems. Unlike other approaches, we prove that TLDR is able to learn the physical parameters of the system, hence enhancing interpretability. We show that our method performs competitively by comparing it to state-of-the-art algorithms on two systems derived from scientific modeling.

Submission Number: 46