Learning Latent Variable Models with Discriminant Regularization

Jing Peng, Alex J. Aved

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


In many machine learning applications, data are often described by a large number of features or attributes. However, too many features can result in overfitting. This is often the case when the number of examples is smaller than the number of features. The problem can be mitigated by learning latent variable models where the data can be described by a fewer number of latent dimensions. There are many techniques for learning latent variable models in the literature. Most of these techniques can be grouped into two classes: techniques that are informative, represented by principal component analysis (PCA), and techniques that are discriminant, represented by linear discriminant analysis (LDA). Each class of the techniques has its advantages. In this work, we introduce a technique for learning latent variable models with discriminant regularization that combines the characteristics of both classes. Empirical evaluation using a variety of data sets is presented to verify the performance of the proposed technique.

Original languageEnglish
Title of host publicationAgents and Artificial Intelligence - 12th International Conference, ICAART 2020, Revised Selected Papers
EditorsAna Paula Rocha, Luc Steels, Jaap van den Herik
PublisherSpringer Science and Business Media Deutschland GmbH
Number of pages21
ISBN (Print)9783030711573
StatePublished - 2021
Event12th International Conference on Agents and Artificial Intelligence, ICAART 2020 - Valletta, Malta
Duration: 22 Feb 202024 Feb 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12613 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference12th International Conference on Agents and Artificial Intelligence, ICAART 2020


  • Classification
  • Dimensionality reduction
  • Latent variable models


Dive into the research topics of 'Learning Latent Variable Models with Discriminant Regularization'. Together they form a unique fingerprint.

Cite this