Deep Learning with Logical Constraints

TL;DR

This survey reviews how integrating logical constraints into deep learning models enhances performance, data efficiency, and safety compliance, categorizing approaches by logical language and objectives.

Contribution

It provides a comprehensive categorization and analysis of recent methods combining logical constraints with deep learning, highlighting their goals and logical frameworks.

Findings

Logical integration improves model performance

Enhances learning from limited data

Ensures safety and compliance in critical applications

Abstract

In recent years, there has been an increasing interest in exploiting logically specified background knowledge in order to obtain neural models (i) with a better performance, (ii) able to learn from less data, and/or (iii) guaranteed to be compliant with the background knowledge itself, e.g., for safety-critical applications. In this survey, we retrace such works and categorize them based on (i) the logical language that they use to express the background knowledge and (ii) the goals that they achieve.