Unsupervised Abstract Reasoning for Raven's Problem Matrices
This addresses the challenge of measuring abstract reasoning in AI for cognitive science and machine learning, though it is incremental as it builds on existing RPM benchmarks.
The paper tackles the problem of solving Raven's Progressive Matrices (RPM) without ground truth labels by proposing the first unsupervised learning method, which outperforms some supervised approaches on three datasets.
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.