Abstract Reasoning with Distracting Features

TL;DR

This paper addresses the challenge of distracting features in abstract reasoning tasks by proposing a novel reinforcement learning-based training strategy and a new model, FRAR, which significantly improves performance on benchmark datasets.

Contribution

The paper introduces the FRAR model with a reinforcement learning teacher to optimize training sequences, effectively reducing the impact of distracting features in abstract reasoning.

Findings

FRAR outperforms baseline algorithms by 18.7% on RAVEN dataset

FRAR outperforms baseline algorithms by 13.3% on PGM dataset

Carefully designed training trajectories improve learning in the presence of distracting features

Abstract

Abstraction reasoning is a long-standing challenge in artificial intelligence. Recent studies suggest that many of the deep architectures that have triumphed over other domains failed to work well in abstract reasoning. In this paper, we first illustrate that one of the main challenges in such a reasoning task is the presence of distracting features, which requires the learning algorithm to leverage counterevidence and to reject any of the false hypotheses in order to learn the true patterns. We later show that carefully designed learning trajectory over different categories of training data can effectively boost learning performance by mitigating the impacts of distracting features. Inspired by this fact, we propose feature robust abstract reasoning (FRAR) model, which consists of a reinforcement learning based teacher network to determine the sequence of training and a student network for predictions. Experimental results demonstrated strong improvements over baseline algorithms and we are able to beat the state-of-the-art models by 18.7% in the RAVEN dataset and 13.3% in the PGM dataset.