One Self-Configurable Model to Solve Many Abstract Visual Reasoning Problems

TL;DR

This paper introduces SCAR, a self-configurable, unified model capable of solving diverse abstract visual reasoning problems without prior task-specific assumptions, advancing towards a general AI reasoning system.

Contribution

The paper presents the first self-configurable, unified model for single-choice AVR tasks that adapts to various problem structures without prior assumptions.

Findings

SCAR performs on par with state-of-the-art task-specific models.

SCAR demonstrates effective multi-task and transfer learning capabilities.

The model successfully solves diverse AVR problems like Raven's Matrices and Odd One Out.

Abstract

Abstract Visual Reasoning (AVR) comprises a wide selection of various problems similar to those used in human IQ tests. Recent years have brought dynamic progress in solving particular AVR tasks, however, in the contemporary literature AVR problems are largely dealt with in isolation, leading to highly specialized task-specific methods. With the aim of developing universal learning systems in the AVR domain, we propose the unified model for solving Single-Choice Abstract visual Reasoning tasks (SCAR), capable of solving various single-choice AVR tasks, without making any a priori assumptions about the task structure, in particular the number and location of panels. The proposed model relies on a novel Structure-Aware dynamic Layer (SAL), which adapts its weights to the structure of the considered AVR problem. Experiments conducted on Raven's Progressive Matrices, Visual Analogy Problems, and Odd One Out problems show that SCAR (SAL-based models, in general) effectively solves diverse AVR tasks, and its performance is on par with the state-of-the-art task-specific baselines. What is more, SCAR demonstrates effective knowledge reuse in multi-task and transfer learning settings. To our knowledge, this work is the first successful attempt to construct a general single-choice AVR solver relying on self-configurable architecture and unified solving method. With this work we aim to stimulate and foster progress on task-independent research paths in the AVR domain, with the long-term goal of development of a general AVR solver.