Lifted Relational Neural Networks

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.

Expanded section on weight learning, added explanation of relationship to convolutional neural networks

Cited by in corpus (8)

• Logical Rule Induction and Theory Learning Using Neural Theorem Proving
• DeepLogic: Towards End-to-End Differentiable Logical Reasoning
• pix2rule: End-to-end Neuro-symbolic Rule Learning
• Extending Answer Set Programs with Neural Networks
• Knowledge-augmented Column Networks: Guiding Deep Learning with Advice
• Learning Invariants through Soft Unification
• Non-Parametric Learning of Lifted Restricted Boltzmann Machines
• Human-Guided Learning of Column Networks: Augmenting Deep Learning with Advice