Deep Non-Monotonic Reasoning for Visual Abstract Reasoning Tasks

TL;DR

This paper introduces a non-monotonic deep learning approach for visual abstract reasoning, inspired by human reasoning, and demonstrates its effectiveness on the challenging RAVEN dataset compared to traditional monotonic models.

Contribution

It proposes a novel non-monotonic reasoning framework for deep learning models tackling visual abstract reasoning tasks.

Findings

Outperforms existing monotonic models on the RAVEN dataset

More effective in strict, challenging experimental settings

Demonstrates the importance of non-monotonic reasoning in visual tasks

Abstract

While achieving unmatched performance on many well-defined tasks, deep learning models have also been used to solve visual abstract reasoning tasks, which are relatively less well-defined, and have been widely used to measure human intelligence. However, current deep models struggle to match human abilities to solve such tasks with minimum data but maximum generalization. One limitation is that current deep learning models work in a monotonic way, i.e., treating different parts of the input in essentially fixed orderings, whereas people repeatedly observe and reason about the different parts of the visual stimuli until the reasoning process converges to a consistent conclusion, i.e., non-monotonic reasoning. This paper proposes a non-monotonic computational approach to solve visual abstract reasoning tasks. In particular, we implemented a deep learning model using this approach and tested it on the RAVEN dataset -- a dataset inspired by the Raven's Progressive Matrices test. Results show that the proposed approach is more effective than existing monotonic deep learning models, under strict experimental settings that represent a difficult variant of the RAVEN dataset problem.