Learning to Transfer: A Foliated Theory

TL;DR

This paper introduces a geometric framework based on foliations to formally describe and analyze transfer learning, aiming to enhance understanding and development of transfer techniques.

Contribution

It provides the first formal, differential geometric foundation for transfer learning, clarifying task relationships and how they can be exploited.

Findings

Framework clarifies task relationships in transfer learning

Provides a mathematical basis for analyzing transfer methods

Enhances understanding of transfer efficiency and generalization

Abstract

Learning to transfer considers learning solutions to tasks in a such way that relevant knowledge can be transferred from known task solutions to new, related tasks. This is important for general learning, as well as for improving the efficiency of the learning process. While techniques for learning to transfer have been studied experimentally, we still lack a foundational description of the problem that exposes what related tasks are, and how relationships between tasks can be exploited constructively. In this work, we introduce a framework using the differential geometric theory of foliations that provides such a foundation.