Adversarial Deep Averaging Networks for Cross-Lingual Sentiment Classification (2016-06-06T00:00:00.000000Z)

TL;DR

An Adversarial Deep Averaging Network (ADAN1) is proposed to transfer the knowledge learned from labeled data on a resource-rich source language to low-resource languages where only unlabeled data exist.

Abstract

In recent years great success has been achieved in sentiment classification for English, thanks in part to the availability of copious annotated resources. Unfortunately, most languages do not enjoy such an abundance of labeled data. To tackle the sentiment classification problem in low-resource languages without adequate annotated data, we propose an Adversarial Deep Averaging Network (ADAN1) to transfer the knowledge learned from labeled data on a resource-rich source language to low-resource languages where only unlabeled data exist. ADAN has two discriminative branches: a sentiment classifier and an adversarial language discriminator. Both branches take input from a shared feature extractor to learn hidden representations that are simultaneously indicative for the classification task and invariant across languages. Experiments on Chinese and Arabic sentiment classification demonstrate that ADAN significantly outperforms state-of-the-art systems.

Authors

Kilian Q. Weinberger

7 papers

Claire Cardie

14 papers

Xilun Chen

7 papers

TL;DR

Abstract

Authors

References64 items

Automatic differentiation in PyTorch

Word Translation Without Parallel Data

Wasserstein Generative Adversarial Networks

Adversarial Training for Unsupervised Bilingual Lexicon Induction

Cross-language Learning with Adversarial Neural Networks

Cross-lingual Distillation for Text Classification

Improved Training of Wasserstein GANs

Adversarial Discriminative Domain Adaptation

Cross-Lingual Sentiment Classification with Bilingual Document Representation Learning

Sentiment Lexicons for Arabic Social Media

An Empirical Study on Sentiment Classification of Chinese Review using Word Embedding

Character-level Convolutional Networks for Text Classification

Effective Approaches to Attention-based Neural Machine Translation

Incorporating Domain and Sentiment Supervision in Representation Learning for Domain Adaptation

Deep Unordered Composition Rivals Syntactic Methods for Text Classification

Bilingual Word Embeddings from Non-Parallel Document-Aligned Data Applied to Bilingual Lexicon Induction

Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks

Bilingual Word Representations with Monolingual Quality in Mind

Domain-Adversarial Training of Neural Networks

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

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

Adam: A Method for Stochastic Optimization

Leveraging Monolingual Data for Crosslingual Compositional Word Representations

Domain-Adversarial Neural Networks

Deep Recursive Neural Networks for Compositionality in Language

BilBOWA: Fast Bilingual Distributed Representations without Word Alignments

Opinion Mining with Deep Recurrent Neural Networks

GloVe: Global Vectors for Word Representation

Unsupervised Domain Adaptation by Backpropagation

Convolutional Neural Networks for Sentence Classification

The Stanford CoreNLP Natural Language Processing Toolkit

Distributed Representations of Sentences and Documents

Distributed Representations of Words and Phrases and their Compositionality

Bilingual Word Embeddings for Phrase-Based Machine Translation

Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank

Marginalized Denoising Autoencoders for Domain Adaptation

Chinese Sentiment Analysis Using Maximum Entropy

Domain Adaptation for Large-Scale Sentiment Classification: A Deep Learning Approach

Joint Bilingual Sentiment Classification with Unlabeled Parallel Corpora

Multilingual Subjectivity: Are More Languages Better?

Word Representations: A Simple and General Method for Semi-Supervised Learning

MultiUN: A Multilingual Corpus from United Nation Documents

Co-Training for Cross-Lingual Sentiment Classification

Domain Adaptation via Transfer Component Analysis

Using Bilingual Knowledge and Ensemble Techniques for Unsupervised Chinese Sentiment Analysis

Multilingual Subjectivity Analysis Using Machine Translation

An empirical study of sentiment analysis for chinese documents

Biographies, Bollywood, Boom-boxes and Blenders: Domain Adaptation for Sentiment Classification

Learning Multilingual Subjective Language via Cross-Lingual Projections

Cross-Lingual Text Categorization

GENERATIVE ADVERSARIAL NETS

How Translation Alters Sentiment

The united nations parallel corpus

Sentiment after Translation: A Case-Study on Arabic Social Media Posts

Dropout: a simple way to prevent neural networks from overfitting

Torch7: A Matlab-like Environment for Machine Learning

Exploring English Lexicon Knowledge for Chinese Sentiment Analysis

Measuring the Averaged Hausdorff Distance to the Pareto Front of a Multi-Objective Optimization Problem

Proceedings of the 23rd International Conference on Computational Linguistics (COLING 2010)

Visualizing Data using t-SNE

Optimal transport: old and new, volume 338

On Hausdorff Distance Measures

JMLR Workshop and Conference Proceedings

Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)

Field of Study

Journal Information

Name

Page

Volume

Venue Information

Name

Type

URL

Alternate Names