IRMA-International.org: Creator of Knowledge
Information Resources Management Association
Advancing the Concepts & Practices of Information Resources Management in Modern Organizations

Class Prediction in Test Sets with Shifted Distributions

Class Prediction in Test Sets with Shifted Distributions
View Sample PDF
Author(s): Óscar Pérez (Universidad Autónoma de Madrid, Spain)and Manuel Sánchez-Montañés (Universidad Autónoma de Madrid, Spain)
Copyright: 2009
Pages: 7
Source title: Encyclopedia of Artificial Intelligence
Source Author(s)/Editor(s): Juan Ramón Rabuñal Dopico (University of A Coruña, Spain), Julian Dorado (University of A Coruña, Spain)and Alejandro Pazos (University of A Coruña, Spain)
DOI: 10.4018/978-1-59904-849-9.ch044

Purchase

View Class Prediction in Test Sets with Shifted Distributions on the publisher's website for pricing and purchasing information.

Abstract

Machine learning has provided powerful algorithms that automatically generate predictive models from experience. One specific technique is supervised learning, where the machine is trained to predict a desired output for each input pattern x. This chapter will focus on classification, that is, supervised learning when the output to predict is a class label. For instance predicting whether a patient in a hospital will develop cancer or not. In this example, the class label c is a variable having two possible values, “cancer” or “no cancer”, and the input pattern x is a vector containing patient data (e.g. age, gender, diet, smoking habits, etc.). In order to construct a proper predictive model, supervised learning methods require a set of examples xi together with their respective labels ci. This dataset is called the “training set”. The constructed model is then used to predict the labels of a set of new cases xj called the “test set”. In the cancer prediction example, this is the phase when the model is used to predict cancer in new patients.

One common assumption in supervised learning algorithms is that the statistical structure of the training and test datasets are the same (Hastie, Tibshirani & Friedman, 2001). That is, the test set is assumed to have the same attribute distribution p(x) and same class distribution p(c|x) as the training set. However, this is not usually the case in real applications due to different reasons. For instance, in many problems the training dataset is obtained in a specific manner that differs from the way the test dataset will be generated later. Moreover, the nature of the problem may evolve in time. These phenomena cause pTr(x, c)  pTest(x, c), which can degrade the performance of the model constructed in training.

Here we present a new algorithm that allows to re-estimate a model constructed in training using the unlabelled test patterns. We show the convergence properties of the algorithm and illustrate its performance with an artificial problem. Finally we demonstrate its strengths in a heart disease diagnosis problem where the training set is taken from a different hospital than the test set.


Related Content

Kamel Mouloudj, Vu Lan Oanh LE, Achouak Bouarar, Ahmed Chemseddine Bouarar, Dachel Martínez Asanza, Mayuri Srivastava. © 2024. 20 pages.
José Eduardo Aleixo, José Luís Reis, Sandrina Francisca Teixeira, Ana Pinto de Lima. © 2024. 52 pages.
Jorge Figueiredo, Isabel Oliveira, Sérgio Silva, Margarida Pocinho, António Cardoso, Manuel Pereira. © 2024. 24 pages.
Fatih Pinarbasi. © 2024. 20 pages.
Stavros Kaperonis. © 2024. 25 pages.
Thomas Rui Mendes, Ana Cristina Antunes. © 2024. 24 pages.
Nuno Geada. © 2024. 12 pages.
Body Bottom