Unsupervised Abstract Reasoning for Raven's Problem Matrices

TL;DR

This paper introduces the first unsupervised deep learning approach for solving Raven's Progressive Matrices, using pseudo targets and negative answers to improve abstract reasoning without labeled data, outperforming some supervised methods.

Contribution

It proposes a novel unsupervised method for RPM problem solving, utilizing pseudo targets, negative answers, and decentralization to enhance reasoning capabilities.

Findings

Outperforms some supervised approaches on three datasets.

Effective pseudo target design for unsupervised RPM solving.

Improved feature adaptation across different RPM problems.

Abstract

Raven's Progressive Matrices (RPM) is highly correlated with human intelligence, and it has been widely used to measure the abstract reasoning ability of humans. In this paper, to study the abstract reasoning capability of deep neural networks, we propose the first unsupervised learning method for solving RPM problems. Since the ground truth labels are not allowed, we design a pseudo target based on the prior constraints of the RPM formulation to approximate the ground truth label, which effectively converts the unsupervised learning strategy into a supervised one. However, the correct answer is wrongly labelled by the pseudo target, and thus the noisy contrast will lead to inaccurate model training. To alleviate this issue, we propose to improve the model performance with negative answers. Moreover, we develop a decentralization method to adapt the feature representation to different RPM problems. Extensive experiments on three datasets demonstrate that our method even outperforms some of the supervised approaches. Our code is available at https://github.com/visiontao/ncd.