Evaluating Relaxations of Logic for Neural Networks: A Comprehensive Study

TL;DR

This paper investigates how different relaxations of logical expressions, particularly t-norms, affect neural network training, providing theoretical insights and empirical results to guide the choice of relaxation methods.

Contribution

It offers a comprehensive analysis of t-norm relaxations for logic in neural networks, combining theoretical criteria with empirical evaluations to identify best practices.

Findings

Lukasiewicz t-norm best preserves tautologies theoretically

Product t-norm yields superior empirical performance

Guidelines for selecting logic relaxations in neural network training

Abstract

Symbolic knowledge can provide crucial inductive bias for training neural models, especially in low data regimes. A successful strategy for incorporating such knowledge involves relaxing logical statements into sub-differentiable losses for optimization. In this paper, we study the question of how best to relax logical expressions that represent labeled examples and knowledge about a problem; we focus on sub-differentiable t-norm relaxations of logic. We present theoretical and empirical criteria for characterizing which relaxation would perform best in various scenarios. In our theoretical study driven by the goal of preserving tautologies, the Lukasiewicz t-norm performs best. However, in our empirical analysis on the text chunking and digit recognition tasks, the product t-norm achieves best predictive performance. We analyze this apparent discrepancy, and conclude with a list of best practices for defining loss functions via logic.