Название: Improving Classifier Generalization: Real-Time Machine Learning based Applications
Автор: Rahul Kumar Sevakula, Nishchal K. Verma
Издательство: Springer
Серия: Studies in Computational Intelligence
Год: 2023
Страниц: 181
Язык: английский
Формат: pdf
Размер: 10.1 MB

This book elaborately discusses techniques commonly used to improve generalization performance in classification approaches. The contents highlight methods to improve classification performance in numerous case studies. The book specifically provides a detailed tutorial on how to approach time-series classification problems and discusses two real time case studies on condition monitoring. In addition to describing the various aspects a data scientist must consider before finalizing their approach to a classification problem and reviewing the state of the art for improving classification generalization performance, it also discusses in detail the authors own contributions to the field, including MVPC - a classifier with very low VC dimension, a graphical indices based framework for reliable predictive maintenance and a novel general-purpose membership functions for Fuzzy Support Vector Machine which provides state of the art performance with noisy datasets, and a novel scheme to introduce Deep Learning (DL) in Fuzzy Rule based classifiers (FRCs).

Classification algorithms form the basis of decision-making in most pattern recognition problems, e.g. image recognition, speech and speaker recognition, iris recognition, and spam mail detection. With the horizon of their applications expanding at a fast pace, the need for further research has only increased. This fact becomes particularly true because (a) each application poses its own set of challenges and (b) one would always find a classifier with a particular improvisation that best suits the situation. No matter which classification approach is used, generalization is an important aspect. Generalization essentially indicates how well the trained classifier works in real time, i.e. on unseen test data.

This monograph begins with the fundamentals of classifiers, bias-variance tradeoff, statistical learning theory (SLT), probably approximate correct (PAC) framework, maximum margin classifiers, and popular methods which improve generalization like regularization, boosting, transfer learning, dropout in Deep Learning, etc. Furthermore, the monograph solves four independent problems that have great relevance for certain real-time applications.

This volume will serve as a useful reference for researchers and students working on Machine Learning, health monitoring, predictive maintenance, time-series analysis, gene-expression data classification.