Neural Symbolic Machines: Learning Semantic Parsers on Freebase with Weak Supervision

Abstract

Harnessing the statistical power of neural networks to perform language understanding and symbolic reasoning is difficult, when it requires executing efficient discrete operations against a large knowledge-base. In this work, we introduce a Neural Symbolic Machine, which contains (a) a neural "programmer", i.e., a sequence-to-sequence model that maps language utterances to programs and utilizes a key-variable memory to handle compositionality (b) a symbolic "computer", i.e., a Lisp interpreter that performs program execution, and helps find good programs by pruning the search space. We apply REINFORCE to directly optimize the task reward of this structured prediction problem. To train with weak supervision and improve the stability of REINFORCE, we augment it with an iterative maximum-likelihood training process. NSM outperforms the state-of-the-art on the WebQuestionsSP dataset when trained from question-answer pairs only, without requiring any feature engineering or domain-specific knowledge.

ACL 2017 camera ready version

References in corpus (6)

• Sequence to Sequence Learning with Neural Networks
• Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
• Learning to Map Sentences to Logical Form: Structured Classification with Probabilistic Categorial Grammars
• Deep Reinforcement Learning for Dialogue Generation
• Sequence-to-Sequence Learning as Beam-Search Optimization
• Reward Augmented Maximum Likelihood for Neural Structured Prediction

Cited by in corpus (28)

• AutoML-Zero: Evolving Machine Learning Algorithms From Scratch
• A Syntactic Neural Model for General-Purpose Code Generation
• DeepPath: A Reinforcement Learning Method for Knowledge Graph Reasoning
• Improved Neural Relation Detection for Knowledge Base Question Answering
• Generative Adversarial Self-Imitation Learning
• Rewriting History with Inverse RL: Hindsight Inference for Policy Improvement
• Recent Advances and Challenges in Task-oriented Dialog System
• Learning Self-Imitating Diverse Policies
• Neural Machine Translation for Query Construction and Composition
• Neural Program Search: Solving Programming Tasks from Description and Examples
• Dynamically Pruned Message Passing Networks for Large-Scale Knowledge Graph Reasoning
• Bidirectional Attentive Memory Networks for Question Answering over Knowledge Bases
• Why Build an Assistant in Minecraft?
• DeepLogic: Towards End-to-End Differentiable Logical Reasoning
• CraftAssist Instruction Parsing: Semantic Parsing for a Minecraft Assistant
• AR-LSAT: Investigating Analytical Reasoning of Text
• Multi-Task Learning for Conversational Question Answering over a Large-Scale Knowledge Base
• Knowledge Base Relation Detection via Multi-View Matching
• Weakly-supervised Neural Semantic Parsing with a Generative Ranker
• Training like Playing: A Reinforcement Learning And Knowledge Graph-based framework for building Automatic Consultation System in Medical Field
• Translating Natural Language to SQL using Pointer-Generator Networks and How Decoding Order Matters
• Generative Question Refinement with Deep Reinforcement Learning in Retrieval-based QA System
• Merging Weak and Active Supervision for Semantic Parsing
• Neural Consciousness Flow
• Logical Parsing from Natural Language Based on a Neural Translation Model
• Learning to Learn Semantic Parsers from Natural Language Supervision
• Context Dependent Semantic Parsing: A Survey
• Differentiable Representations For Multihop Inference Rules