Abstract:Neuro-symbolic models of artificial intelligence (AI) have been recently developed to perform tasks involving abstract visual reasoning that is a hallmark of human intelligence but remains challenging for deep neural network methods. However, most of the current neuro-symbolic models also rely on supervised learning and auxiliary annotations, different from human cognitive processes that are much dependent on the general cognitive abilities of entity and rule recognitions, rather than learning how to solve the specific tasks from examples. In this work, we propose a neuro-symbolic model by self-supervised contrastive learning (NS-SSCL) with unique and invariant representations of entities and rules in the perception and reasoning modules, respectively, to solve Raven’s Progressive Matrices (RPMs) and its variant, a typical type of visual reasoning task used to test human intelligence. The perception module parses each object into invariant representations of attributes. The reasoning module grounds the representations of object attributes to form the latent rule representations also through SSCL. Further, the relationships between the neural representations of object attributes and symbols used for rule reasoning are coherently mapped. Finally, the scene generation engine aggregates all attribute and rule representation distributions to produce a probabilistic representation of the target. NS-SSCL obtains state-of-the-art performance in unsupervised models to solve the RAVEN and V-PROM benchmarks, even better than most of the supervised models. The success of the proposed model suggests that construction of general cognitive abilities like humans may render the AI algorithms to solve complex tasks involving higher-level cognition such as abstract reasoning in a human-like manner.

Learning Visual Abstract Reasoning through Dual-Stream Networks

Learning the Dynamics of Visual Relational Reasoning Via Reinforced Path Routing

Learning to reason over visual objects

A Cognitively-Inspired Neural Architecture for Visual Abstract Reasoning Using Contrastive Perceptual and Conceptual Processing

Multi-Granularity Modularized Network for Abstract Visual Reasoning

Deep Residual Networks’ Abstract Reasoning Performance on Raven’s Progressive Matrices

DAReN: A Collaborative Approach Towards Reasoning And Disentangling

Toward Accurate Visual Reasoning with Dual-Path Neural Module Networks.

Stratified Rule-Aware Network for Abstract Visual Reasoning

Abstract Visual Reasoning by Self-supervised Contrastive Learning

ViTCN: Vision Transformer Contrastive Network For Reasoning

Neural Prediction Errors Enable Analogical Visual Reasoning in Human Standard Intelligence Tests.

Hierarchical Rule Induction Network for Abstract Visual Reasoning

Abstract Reasoning with Distracting Features

RAVEN: A Dataset for Relational and Analogical Visual rEasoNing

A Feature-based Generalizable Prediction Model for Both Perceptual and Abstract Reasoning

Towards Learning Implicit Symbolic Representation for Visual Reasoning

Towards A Unified Neural Architecture for Visual Recognition and Reasoning

Abstract Spatial-Temporal Reasoning Via Probabilistic Abduction and Execution

Learning to Reason Iteratively and Parallelly for Complex Visual Reasoning Scenarios

Deep Neural Networks for Visual Reasoning