Hybrid Classifiers

Methods of Data, Knowledge, and Classifier Combination

verfasst von: Michal Wozniak

Verlag: Springer Berlin Heidelberg

Buchreihe : Studies in Computational Intelligence

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

This book delivers a definite and compact knowledge on how hybridization can help improving the quality of computer classification systems. In order to make readers clearly realize the knowledge of hybridization, this book primarily focuses on introducing the different levels of hybridization and illuminating what problems we will face with as dealing with such projects. In the first instance the data and knowledge incorporated in hybridization were the action points, and then a still growing up area of classifier systems known as combined classifiers was considered. This book comprises the aforementioned state-of-the-art topics and the latest research results of the author and his team from Department of Systems and Computer Networks, Wroclaw University of Technology, including as classifier based on feature space splitting, one-class classification, imbalance data, and data stream classification.

Inhaltsverzeichnis

Frontmatter

Introduction

Abstract

This chapter introduces the idea of machine learning for pattern recognition tasks, describes its main stages, and focuses on classification problems, and then the selected classifiers are presented and discussed in the consequent part. Moreover, the main principles of classifier evaluation are included in this chapter as well, and on the end the definition of the hybrid classifier is proposed.

Michał Woźniak

Data and Knowledge Hybridization

Abstract

This chapter presents the selected issues related to the data and knowledge hybridization, and we focus on the problem of knowledge and data consistency, selected aspects of cost-sensitive, as well as privacy preserving classification. Further, we also present an idea of data and knowledge unification which helps to unify form of the learning material.

Michał Woźniak

Classifier Hybridization

Abstract

Another possibility of hybridization is using the group of classifiers to make a common decision. This chapter introduces this topic and presents a short introduction of main components of combined classifiers, such as topology, ensemble forming, and combination rule. Several methods of such classifier design, learning, and evaluation will be presented as well.

Michał Woźniak

Chosen Applications of Hybrid Classifiers

Abstract

In this chapter, we focus on the chosen methods of classifier hybridization. Firstly, we present a special case of static classifier selection approaches leading to the combined classifier based on feature space partitioning and assigning a chosen classifier to each partition. Meanwhile, we discuss how to train it and then shortly discuss its quality. Afterwards, we concentrate on a case of binary classification tasks called one-class classification, which is able to train a classifier in the absence of counterexamples, and we trash the problem out by considering how to produce combined classifier for multi-class and one class classification tasks. Next, an important topic devoted to classification systems for imbalanced data is mentioned. Also, we shortly mention the last topic related to the data stream classification which nowadays seems to be a crucial classification task. Finally, We introduce the problem about how to employ methods presented in the previous chapters, to the classification task where the data probability characteristics are changing during classifier exploitation. This phenomena, called concept drift, has usually a negative impact on classification quality.

Michał Woźniak

Conclusions

Abstract

The main aims of this book was to deliver a either definite or compact knowledge on how hybridization can help improving the quality of computer classification systems. In order to make readers clearly realize the knowledge of hybridization, this book primarily focuses on introducing the different levels of hybridization and illuminating what problems we will face with as dealing with such projects. In the first instance the data and knowledge incorporated in hybridization were the action points, and then a still growing up area of classifier systems known as combined classifiers was considered.

Michał Woźniak

Backmatter

Titel: Hybrid Classifiers
verfasst von: Michal Wozniak
Verlag: Springer Berlin Heidelberg
Electronic ISBN: 978-3-642-40997-4
Print ISBN: 978-3-642-40996-7
DOI: https://doi.org/10.1007/978-3-642-40997-4