On the Combination of Logistic Regression and Local Probability Estimates

Melanie Osl

Abstract


In this paper we give a survey of the combination of classifiers. We briefly describe basic principles of machine learning and the problem of classifier construction and review several approaches to generate different classifiers as well as established methods to combine different classifiers. Then, we introduce our novel approach to assess the appropriateness of different classifiers based on their characteristics for each test point individually

Keywords


classifier characteristics, classifier combination, ensemble classification, k-nearest neighbor classification, logistic regression, machine learning

Full Text:

PDF PDF