Disentangled Variational Auto-Encoder for semi-supervised learning

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

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Semi-supervised learning is attracting increasing attention due to the fact that datasets of many domains lack enough labeled data. Variational Auto-Encoder (VAE), in particular, has demonstrated the benefits of semi-supervised learning. The majority of existing semi-supervised VAEs utilize a classifier to exploit label information, where the parameters of the classifier are introduced to the VAE. Given the limited labeled data, learning the parameters for the classifiers may not be an optimal solution for exploiting label information. Therefore, in this paper, we develop a novel approach for semi-supervised VAE without classifier. Specifically, we propose a new model called Semi-supervised Disentangled VAE (SDVAE), which encodes the input data into disentangled representation and non-interpretable representation, then the category information is directly utilized to regularize the disentangled representation via the equality constraint. To further enhance the feature learning ability of the proposed VAE, we incorporate reinforcement learning to relieve the lack of data. The dynamic framework is capable of dealing with both image and text data with its corresponding encoder and decoder networks. Extensive experiments on image and text datasets demonstrate the effectiveness of the proposed framework.

Original languageEnglish (US)
Pages (from-to)73-85
Number of pages13
JournalInformation Sciences
Volume482
DOIs
StatePublished - May 1 2019

Fingerprint

Semi-supervised Learning
Supervised learning
Encoder
Classifiers
Classifier
Labels
Reinforcement learning
Equality Constraints
Reinforcement Learning
Semi-supervised learning
Optimal Solution
Experiments
Demonstrate
Experiment

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 ; Peng, Haiyun ; Yang, Tao ; Cambria, Erik. / Disentangled Variational Auto-Encoder for semi-supervised learning. In: Information Sciences. 2019 ; Vol. 482. pp. 73-85.
@article{3618ad9c8d0f4881afc24b3c29556198,
title = "Disentangled Variational Auto-Encoder for semi-supervised learning",
abstract = "Semi-supervised learning is attracting increasing attention due to the fact that datasets of many domains lack enough labeled data. Variational Auto-Encoder (VAE), in particular, has demonstrated the benefits of semi-supervised learning. The majority of existing semi-supervised VAEs utilize a classifier to exploit label information, where the parameters of the classifier are introduced to the VAE. Given the limited labeled data, learning the parameters for the classifiers may not be an optimal solution for exploiting label information. Therefore, in this paper, we develop a novel approach for semi-supervised VAE without classifier. Specifically, we propose a new model called Semi-supervised Disentangled VAE (SDVAE), which encodes the input data into disentangled representation and non-interpretable representation, then the category information is directly utilized to regularize the disentangled representation via the equality constraint. To further enhance the feature learning ability of the proposed VAE, we incorporate reinforcement learning to relieve the lack of data. The dynamic framework is capable of dealing with both image and text data with its corresponding encoder and decoder networks. Extensive experiments on image and text datasets demonstrate the effectiveness of the proposed framework.",
author = "Yang Li and Quan Pan and Suhang Wang and Haiyun Peng and Tao Yang and Erik Cambria",
year = "2019",
month = "5",
day = "1",
doi = "10.1016/j.ins.2018.12.057",
language = "English (US)",
volume = "482",
pages = "73--85",
journal = "Information Sciences",
issn = "0020-0255",
publisher = "Elsevier Inc.",

}

Disentangled Variational Auto-Encoder for semi-supervised learning. / Li, Yang; Pan, Quan; Wang, Suhang; Peng, Haiyun; Yang, Tao; Cambria, Erik.

In: Information Sciences, Vol. 482, 01.05.2019, p. 73-85.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Disentangled Variational Auto-Encoder for semi-supervised learning

AU - Li, Yang

AU - Pan, Quan

AU - Wang, Suhang

AU - Peng, Haiyun

AU - Yang, Tao

AU - Cambria, Erik

PY - 2019/5/1

Y1 - 2019/5/1

N2 - Semi-supervised learning is attracting increasing attention due to the fact that datasets of many domains lack enough labeled data. Variational Auto-Encoder (VAE), in particular, has demonstrated the benefits of semi-supervised learning. The majority of existing semi-supervised VAEs utilize a classifier to exploit label information, where the parameters of the classifier are introduced to the VAE. Given the limited labeled data, learning the parameters for the classifiers may not be an optimal solution for exploiting label information. Therefore, in this paper, we develop a novel approach for semi-supervised VAE without classifier. Specifically, we propose a new model called Semi-supervised Disentangled VAE (SDVAE), which encodes the input data into disentangled representation and non-interpretable representation, then the category information is directly utilized to regularize the disentangled representation via the equality constraint. To further enhance the feature learning ability of the proposed VAE, we incorporate reinforcement learning to relieve the lack of data. The dynamic framework is capable of dealing with both image and text data with its corresponding encoder and decoder networks. Extensive experiments on image and text datasets demonstrate the effectiveness of the proposed framework.

AB - Semi-supervised learning is attracting increasing attention due to the fact that datasets of many domains lack enough labeled data. Variational Auto-Encoder (VAE), in particular, has demonstrated the benefits of semi-supervised learning. The majority of existing semi-supervised VAEs utilize a classifier to exploit label information, where the parameters of the classifier are introduced to the VAE. Given the limited labeled data, learning the parameters for the classifiers may not be an optimal solution for exploiting label information. Therefore, in this paper, we develop a novel approach for semi-supervised VAE without classifier. Specifically, we propose a new model called Semi-supervised Disentangled VAE (SDVAE), which encodes the input data into disentangled representation and non-interpretable representation, then the category information is directly utilized to regularize the disentangled representation via the equality constraint. To further enhance the feature learning ability of the proposed VAE, we incorporate reinforcement learning to relieve the lack of data. The dynamic framework is capable of dealing with both image and text data with its corresponding encoder and decoder networks. Extensive experiments on image and text datasets demonstrate the effectiveness of the proposed framework.

UR - http://www.scopus.com/inward/record.url?scp=85059744324&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85059744324&partnerID=8YFLogxK

U2 - 10.1016/j.ins.2018.12.057

DO - 10.1016/j.ins.2018.12.057

M3 - Article

AN - SCOPUS:85059744324

VL - 482

SP - 73

EP - 85

JO - Information Sciences

JF - Information Sciences

SN - 0020-0255

ER -