Selective Replay Enhances Learning in Online Continual Analogical Reasoning

TL;DR

This paper explores continual learning in neural networks for abstract reasoning, demonstrating that selective experience replay significantly improves learning performance on Raven's Progressive Matrices compared to random replay.

Contribution

It introduces experimental protocols and metrics for continual analogical reasoning, and shows that selective replay enhances learning in RPM tasks, unlike in image classification.

Findings

Selective replay outperforms random replay in RPM tasks

Established benchmarks and metrics for continual analogical reasoning

Demonstrated effectiveness of replay strategies in abstract reasoning

Abstract

In continual learning, a system learns from non-stationary data streams or batches without catastrophic forgetting. While this problem has been heavily studied in supervised image classification and reinforcement learning, continual learning in neural networks designed for abstract reasoning has not yet been studied. Here, we study continual learning of analogical reasoning. Analogical reasoning tests such as Raven's Progressive Matrices (RPMs) are commonly used to measure non-verbal abstract reasoning in humans, and recently offline neural networks for the RPM problem have been proposed. In this paper, we establish experimental baselines, protocols, and forward and backward transfer metrics to evaluate continual learners on RPMs. We employ experience replay to mitigate catastrophic forgetting. Prior work using replay for image classification tasks has found that selectively choosing the samples to replay offers little, if any, benefit over random selection. In contrast, we find that selective replay can significantly outperform random selection for the RPM task.