Can Neural Networks Understand Logical Entailment?

2018/02/23 by David Saxton, Evans, Richard, Saxton, David +6 · 2 citations
Computer Science · #Artificial Intelligence (cs.AI) #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Natural Language Processing Techniques #Neural and Evolutionary Computing (cs.NE) #Topic Modeling

paper · pdf · doi:10.48550/arxiv.1802.08535

openalex publication_date 2018/02/23 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We introduce a new dataset of logical entailments for the purpose of measuring models' ability to capture and exploit the structure of logical expressions against an entailment prediction task. We use this task to compare a series of architectures which are ubiquitous in the sequence-processing literature, in addition to a new model class---PossibleWorldNets---which computes entailment as a "convolution over possible worlds". Results show that convolutional networks present the wrong inductive bias for this class of problems relative to LSTM RNNs, tree-structured neural networks outperform LSTM RNNs due to their enhanced ability to exploit the syntax of logic, and PossibleWorldNets outperform all benchmarks.

Citations

• Adam: A Method for Stochastic Optimization
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Neural Machine Translation by Jointly Learning to Align and Translate
• Learning to Discover Efficient Mathematical Identities
• Better Computer Go Player with Neural Network and Long-term Prediction
• Neural GPUs Learn Algorithms
• Neural Programmer-Interpreters
• Latent Predictor Networks for Code Generation
• A large annotated corpus for learning natural language inference
• ABCNN: Attention-Based Convolutional Neural Network for Modeling Sentence Pairs

Cited by

• Zero-Shot Instruction Following in RL via Structured LTL Representations
• Natural Language Satisfiability: Exploring the Problem Distribution and Evaluating Transformer-based Language Models

Related

• Siamese recurrent networks learn first-order logic reasoning and exhibit zero-shot compositional generalization
• Recursive Neural Networks Can Learn Logical Semantics
• A Logical Pattern Memory Pre-trained Model for Entailment Tree Generation
• Reasoning about Entailment with Neural Attention
• Neural Logic Machines