Measuring the Feasibility of Analogical Transfer using Complexity

TL;DR

This paper introduces a method to quantify the feasibility of solving analogies based on complexity minimization, connecting analogy transferability with machine learning and domain adaptation.

Contribution

It proposes a novel complexity-based measure for assessing the transferability of analogies, with applications to morphological analogies and links to domain adaptation.

Findings

The complexity measure effectively predicts analogy transferability.

Connections established between analogy solving and machine learning techniques.

Illustrations on morphological analogies demonstrate practical relevance.

Abstract

Analogies are 4-ary relations of the form "A is to B as C is to D". While focus has been mostly on how to solve an analogy, i.e. how to find correct values of D given A, B and C, less attention has been drawn on whether solving such an analogy was actually feasible. In this paper, we propose a quantification of the transferability of a source case (A and B) to solve a target problem C. This quantification is based on a complexity minimization principle which has been demonstrated to be efficient for solving analogies. We illustrate these notions on morphological analogies and show its connections with machine learning, and in particular with Unsupervised Domain Adaptation.