A Generative Model for category text generation

Yang Li, Quan Pan, Suhang Wang, Tao Yang, Erik Cambria

Research output: Contribution to journalArticle

16 Citations (Scopus)

Abstract

The neural network model has been the fulcrum of the so-called AI revolution. Although very powerful for pattern-recognition tasks, however, the model has two main drawbacks: it tends to overfit when the training dataset is small, and it is unable to accurately capture category information when the class number is large. In this paper, we combine reinforcement learning, generative adversarial networks, and recurrent neural networks to build a new model, termed category sentence generative adversarial network (CS-GAN). Not only the proposed model is able to generate category sentences that enlarge the original dataset, but also it helps improve its generalization capability during supervised training. We evaluate the performance of CS-GAN for the task of sentiment analysis. Quantitative evaluation exhibits the accuracy improvement in polarity detection on a small dataset with high category information.

Original languageEnglish (US)
Pages (from-to)301-315
Number of pages15
JournalInformation Sciences
Volume450
DOIs
StatePublished - Jun 1 2018

Fingerprint

Generative Models
Fulcrum
Sentiment Analysis
Model Category
Recurrent neural networks
Quantitative Evaluation
Class number
Reinforcement learning
Recurrent Neural Networks
Polarity
Reinforcement Learning
Neural Network Model
Pattern Recognition
Pattern recognition
Tend
Neural networks
Text
Evaluate
Model
Training

All Science Journal Classification (ASJC) codes

  • Software
  • Control and Systems Engineering
  • Theoretical Computer Science
  • Computer Science Applications
  • Information Systems and Management
  • Artificial Intelligence

Cite this

Li, Yang ; Pan, Quan ; Wang, Suhang ; Yang, Tao ; Cambria, Erik. / A Generative Model for category text generation. In: Information Sciences. 2018 ; Vol. 450. pp. 301-315.
@article{0797afb62a63460ab18d43f25795f927,
title = "A Generative Model for category text generation",
abstract = "The neural network model has been the fulcrum of the so-called AI revolution. Although very powerful for pattern-recognition tasks, however, the model has two main drawbacks: it tends to overfit when the training dataset is small, and it is unable to accurately capture category information when the class number is large. In this paper, we combine reinforcement learning, generative adversarial networks, and recurrent neural networks to build a new model, termed category sentence generative adversarial network (CS-GAN). Not only the proposed model is able to generate category sentences that enlarge the original dataset, but also it helps improve its generalization capability during supervised training. We evaluate the performance of CS-GAN for the task of sentiment analysis. Quantitative evaluation exhibits the accuracy improvement in polarity detection on a small dataset with high category information.",
author = "Yang Li and Quan Pan and Suhang Wang and Tao Yang and Erik Cambria",
year = "2018",
month = "6",
day = "1",
doi = "10.1016/j.ins.2018.03.050",
language = "English (US)",
volume = "450",
pages = "301--315",
journal = "Information Sciences",
issn = "0020-0255",
publisher = "Elsevier Inc.",

}

A Generative Model for category text generation. / Li, Yang; Pan, Quan; Wang, Suhang; Yang, Tao; Cambria, Erik.

In: Information Sciences, Vol. 450, 01.06.2018, p. 301-315.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A Generative Model for category text generation

AU - Li, Yang

AU - Pan, Quan

AU - Wang, Suhang

AU - Yang, Tao

AU - Cambria, Erik

PY - 2018/6/1

Y1 - 2018/6/1

N2 - The neural network model has been the fulcrum of the so-called AI revolution. Although very powerful for pattern-recognition tasks, however, the model has two main drawbacks: it tends to overfit when the training dataset is small, and it is unable to accurately capture category information when the class number is large. In this paper, we combine reinforcement learning, generative adversarial networks, and recurrent neural networks to build a new model, termed category sentence generative adversarial network (CS-GAN). Not only the proposed model is able to generate category sentences that enlarge the original dataset, but also it helps improve its generalization capability during supervised training. We evaluate the performance of CS-GAN for the task of sentiment analysis. Quantitative evaluation exhibits the accuracy improvement in polarity detection on a small dataset with high category information.

AB - The neural network model has been the fulcrum of the so-called AI revolution. Although very powerful for pattern-recognition tasks, however, the model has two main drawbacks: it tends to overfit when the training dataset is small, and it is unable to accurately capture category information when the class number is large. In this paper, we combine reinforcement learning, generative adversarial networks, and recurrent neural networks to build a new model, termed category sentence generative adversarial network (CS-GAN). Not only the proposed model is able to generate category sentences that enlarge the original dataset, but also it helps improve its generalization capability during supervised training. We evaluate the performance of CS-GAN for the task of sentiment analysis. Quantitative evaluation exhibits the accuracy improvement in polarity detection on a small dataset with high category information.

UR - http://www.scopus.com/inward/record.url?scp=85044670076&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85044670076&partnerID=8YFLogxK

U2 - 10.1016/j.ins.2018.03.050

DO - 10.1016/j.ins.2018.03.050

M3 - Article

AN - SCOPUS:85044670076

VL - 450

SP - 301

EP - 315

JO - Information Sciences

JF - Information Sciences

SN - 0020-0255

ER -