Differentiable Learning of Logical Rules for Knowledge Base Reasoning

Abstract

We study the problem of learning probabilistic first-order logical rules for knowledge base reasoning. This learning problem is difficult because it requires learning the parameters in a continuous space as well as the structure in a discrete space. We propose a framework, Neural Logic Programming, that combines the parameter and structure learning of first-order logical rules in an end-to-end differentiable model. This approach is inspired by a recently-developed differentiable logic called TensorLog, where inference tasks can be compiled into sequences of differentiable operations. We design a neural controller system that learns to compose these operations. Empirically, our method outperforms prior work on multiple knowledge base benchmark datasets, including Freebase and WikiMovies.

Accepted at NIPS 2017

References in corpus (5)

• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
• Question Answering with Subgraph Embeddings
• TensorLog: A Differentiable Deductive Database
• Words or Characters? Fine-grained Gating for Reading Comprehension
• Link Prediction using Embedded Knowledge Graphs

Cited by in corpus (30)

• Knowledge Graphs
• A Novel Embedding Model for Knowledge Base Completion Based on Convolutional Neural Network
• The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence
• A Boxology of Design Patterns for Hybrid Learning and Reasoning Systems
• Introduction to Neural Network based Approaches for Question Answering over Knowledge Graphs
• Multi-Hop Knowledge Graph Reasoning with Reward Shaping
• Bridging the Gap between Spatial and Spectral Domains: A Survey on Graph Neural Networks
• TensorLog: Deep Learning Meets Probabilistic DBs
• Iteratively Learning Embeddings and Rules for Knowledge Graph Reasoning
• Beyond Graph Neural Networks with Lifted Relational Neural Networks
• Logical Rule Induction and Theory Learning Using Neural Theorem Proving
• Evaluating Logical Generalization in Graph Neural Networks
• Logical Reasoning with Relation Network for Inductive Knowledge Graph Completion
• Learning Representations of Bi-level Knowledge Graphs for Reasoning beyond Link Prediction
• VQA-LOL: Visual Question Answering under the Lens of Logic
• Recurrent Skipping Networks for Entity Alignment
• Concept2vec: Metrics for Evaluating Quality of Embeddings for Ontological Concepts
• Towards Learning Instantiated Logical Rules from Knowledge Graphs
• Learning First-Order Rules with Relational Path Contrast for Inductive Relation Reasoning
• Assessing SATNet's Ability to Solve the Symbol Grounding Problem
• Neural Network for Weighted Signal Temporal Logic
• Dynamic Anticipation and Completion for Multi-Hop Reasoning over Sparse Knowledge Graph
• From Shallow to Deep Interactions Between Knowledge Representation, Reasoning and Machine Learning (Kay R. Amel group)
• Explainable Biomedical Recommendations via Reinforcement Learning Reasoning on Knowledge Graphs
• An Empirical Analysis of Multiple-Turn Reasoning Strategies in Reading Comprehension Tasks
• Combining Rules and Embeddings via Neuro-Symbolic AI for Knowledge Base Completion
• Leveraging Declarative Knowledge in Text and First-Order Logic for Fine-Grained Propaganda Detection
• AI Reasoning Systems: PAC and Applied Methods
• Ontology-based Adaptive e-Textbook Platform for Student and Machine Co-Learning
• Neuro-Symbolic Forward Reasoning