The Scattering Compositional Learner: Discovering Objects, Attributes, Relationships in Analogical Reasoning

TL;DR

This paper introduces the Scattering Compositional Learner (SCL), a neural network architecture that effectively discovers compositional structures in analogical reasoning tasks like Raven's Progressive Matrices, achieving state-of-the-art results and improved robustness.

Contribution

The paper presents SCL, a novel neural network architecture that composes networks sequentially to discover objects, attributes, and relationships in analogical reasoning tasks, enhancing performance and generalization.

Findings

Achieves 48.7% and 26.4% relative improvements on RPM datasets.

Discovers compositional representations of objects and relationships.

Improves robustness to domain shifts and zero-shot generalization.

Abstract

In this work, we focus on an analogical reasoning task that contains rich compositional structures, Raven's Progressive Matrices (RPM). To discover compositional structures of the data, we propose the Scattering Compositional Learner (SCL), an architecture that composes neural networks in a sequence. Our SCL achieves state-of-the-art performance on two RPM datasets, with a 48.7% relative improvement on Balanced-RAVEN and 26.4% on PGM over the previous state-of-the-art. We additionally show that our model discovers compositional representations of objects' attributes (e.g., shape color, size), and their relationships (e.g., progression, union). We also find that the compositional representation makes the SCL significantly more robust to test-time domain shifts and greatly improves zero-shot generalization to previously unseen analogies.