Measuring abstract reasoning in neural networks
This work addresses the debate on whether neural networks can learn abstract reasoning, providing a benchmark for the AI research community, though it is incremental in exploring specific generalization regimes.
The authors tackled the problem of measuring abstract reasoning in neural networks by introducing a dataset inspired by human IQ tests, showing that popular models like ResNets perform poorly while their novel architecture achieves significantly better results, with improvements when trained to predict symbolic explanations.
Whether neural networks can learn abstract reasoning or whether they merely rely on superficial statistics is a topic of recent debate. Here, we propose a dataset and challenge designed to probe abstract reasoning, inspired by a well-known human IQ test. To succeed at this challenge, models must cope with various generalisation `regimes' in which the training and test data differ in clearly-defined ways. We show that popular models such as ResNets perform poorly, even when the training and test sets differ only minimally, and we present a novel architecture, with a structure designed to encourage reasoning, that does significantly better. When we vary the way in which the test questions and training data differ, we find that our model is notably proficient at certain forms of generalisation, but notably weak at others. We further show that the model's ability to generalise improves markedly if it is trained to predict symbolic explanations for its answers. Altogether, we introduce and explore ways to both measure and induce stronger abstract reasoning in neural networks. Our freely-available dataset should motivate further progress in this direction.