Graph2Seq: Graph to Sequence Learning with Attention-based Neural Networks (2018-04-03T00:00:00.000000Z)

TL;DR

This work introduces a novel general end-to-end graph- to-sequence neural encoder-decoder model that maps an input graph to a sequence of vectors and uses an attention-based LSTM method to decode the target sequence from these vectors.

Abstract

The celebrated Sequence to Sequence learning (Seq2Seq) technique and its numerous variants achieve excellent performance on many tasks. However, many machine learning tasks have inputs naturally represented as graphs; existing Seq2Seq models face a significant challenge in achieving accurate conversion from graph form to the appropriate sequence. To address this challenge, we introduce a novel general end-to-end graph-to-sequence neural encoder-decoder model that maps an input graph to a sequence of vectors and uses an attention-based LSTM method to decode the target sequence from these vectors. Our method first generates the node and graph embeddings using an improved graph-based neural network with a novel aggregation strategy to incorporate edge direction information in the node embeddings. We further introduce an attention mechanism that aligns node embeddings and the decoding sequence to better cope with large graphs. Experimental results on bAbI, Shortest Path, and Natural Language Generation tasks demonstrate that our model achieves state-of-the-art performance and significantly outperforms existing graph neural networks, Seq2Seq, and Tree2Seq models; using the proposed bi-directional node embedding aggregation strategy, the model can converge rapidly to the optimal performance.

Authors

Yansong Feng

4 papers

Lingfei Wu

11 papers

Zhiguo Wang

6 papers

TL;DR

Abstract

Authors

References70 items

Graph-to-Sequence Learning using Gated Graph Neural Networks

TypeSQL: Knowledge-Based Type-Aware Neural Text-to-SQL Generation

Learning Deep Generative Models of Graphs

GraphVAE: Towards Generation of Small Graphs Using Variational Autoencoders

FastGCN: Fast Learning with Graph Convolutional Networks via Importance Sampling

Protein Interface Prediction using Graph Convolutional Networks

Graph Attention Networks

Learning Graph Representations with Embedding Propagation

Representation Learning on Graphs: Methods and Applications

Content Selection for Real-time Sports News Construction from Commentary Texts

Seq2SQL: Generating Structured Queries from Natural Language using Reinforcement Learning

Seq2seq Fingerprint: An Unsupervised Deep Molecular Embedding for Drug Discovery

Inductive Representation Learning on Large Graphs

Retrosynthetic Reaction Prediction Using Neural Sequence-to-Sequence Models

Graph Embedding Techniques, Applications, and Performance: A Survey

Convolutional Sequence to Sequence Learning

Graph Convolutional Encoders for Syntax-aware Neural Machine Translation

Modeling Relational Data with Graph Convolutional Networks

AMR-to-text Generation with Synchronous Node Replacement Grammar

Very deep convolutional networks for end-to-end speech recognition

Automatic Chemical Design Using a Data-Driven Continuous Representation of Molecules

Semi-Supervised Classification with Graph Convolutional Networks

Asymmetric Transitivity Preserving Graph Embedding

Structural Deep Network Embedding

Summarizing Source Code using a Neural Attention Model

node2vec: Scalable Feature Learning for Networks

Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

DeepMath - Deep Sequence Models for Premise Selection

Learning Convolutional Neural Networks for Graphs

Revisiting Semi-Supervised Learning with Graph Embeddings

Incorporating Copying Mechanism in Sequence-to-Sequence Learning

Tree-to-Sequence Attentional Neural Machine Translation

Order Matters: Sequence to sequence for sets

Gated Graph Sequence Neural Networks

GraRep: Learning Graph Representations with Global Structural Information

Convolutional Networks on Graphs for Learning Molecular Fingerprints

Parsing English into Abstract Meaning Representation Using Syntax-Based Machine Translation

Effective Approaches to Attention-based Neural Machine Translation

Attention-Based Models for Speech Recognition

Pointer Networks

End-To-End Memory Networks

LINE: Large-scale Information Network Embedding

Improved Semantic Representations From Tree-Structured Long Short-Term Memory Networks

Towards AI-Complete Question Answering: A Set of Prerequisite Toy Tasks

Adam: A Method for Stochastic Optimization

Show and tell: A neural image caption generator

Learning to Execute

Sequence to Sequence Learning with Neural Networks

Neural Machine Translation by Jointly Learning to Align and Translate

Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation

DeepWalk: online learning of social representations

Spectral Networks and Locally Connected Networks on Graphs

Abstract Meaning Representation for Sembanking

Distributed large-scale natural graph factorization

Deep Sparse Rectifier Neural Networks

A new model for learning in graph domains

2005 Special Issue: Framewise phoneme classification with bidirectional LSTM and other neural network architectures

A knowledge based genetic algorithm for path planning of a mobile robot

Bleu: a Method for Automatic Evaluation of Machine Translation

Laplacian Eigenmaps and Spectral Techniques for Embedding and Clustering

Nonlinear dimensionality reduction by locally linear embedding.

Long Short-Term Memory

Bidirectional recurrent neural networks

Translation of SQL queries into a graph structure: query transformations and pre-optimization issues in a pipeline multiprocessor environment

Dropout: a simple way to prevent neural networks from overfitting

Learning Continuous Phrase Representations and Syntactic Parsing with Recursive Neural Networks

The Graph Neural Network Model

We update the backward representation of v , h kv (cid:97) , using the similar procedure as introduced in step (3) and (4) except that

We concatenate v ’s current forward representation, h k − 1 v (cid:96) , with the newly generated neighborhood vector, h k N (cid:96) ( v ) .

We categorize the neighbors of v into forward neighbors, N

Field of Study

Journal Information

Name

Volume

Venue Information

Name