Lifted Relational Neural Networks

TL;DR

This paper introduces a novel neural network architecture that integrates relational logic with deep learning, enabling hierarchical relational modeling and learning of latent concepts, demonstrated on multiple benchmarks.

Contribution

It presents a lifted relational neural network framework that combines relational rules with neural network structures, allowing for domain knowledge incorporation and shared weight learning.

Findings

Favorable performance on 78 relational learning benchmarks

Ability to discover notable relational concepts

Supports hierarchical relational modeling

Abstract

We propose a method combining relational-logic representations with neural network learning. A general lifted architecture, possibly reflecting some background domain knowledge, is described through relational rules which may be handcrafted or learned. The relational rule-set serves as a template for unfolding possibly deep neural networks whose structures also reflect the structures of given training or testing relational examples. Different networks corresponding to different examples share their weights, which co-evolve during training by stochastic gradient descent algorithm. The framework allows for hierarchical relational modeling constructs and learning of latent relational concepts through shared hidden layers weights corresponding to the rules. Discovery of notable relational concepts and experiments on 78 relational learning benchmarks demonstrate favorable performance of the method.