Publisher's Synopsis
The interface between statistics and machine learning (ML) is an increasingly popular research subject, as scientists and statisticians discover useful areas where these disciplines overlap. ML deals primarily with learning rules or structure, and while some books on the subject exist, this volume is the only one to integrate ML with statistics. It explores new areas where theory and methods can be shared and demonstrates the benefits to those working in either discipline.
Written by leading experts in both fields, Machine Learning and Statistics is a result of the authors′ participation in the 1994 European Conference in Machine Learning. This important collection of contributions was adapted from conference workshop material and reworked to address readers of diverse backgrounds and skills. For newcomers to the field, a thorough introduction surveys the various topics and supplies numerous references for further reading.
The book′s main focus is on classification, the most common area of intersection. The classification process uses information about a new example to assign the example to one of a known number of classes. Such methods typically involve a rule learned from an initial set of data, which is where ML comes into play. Other topics covered include prediction, control, and an introduction to methods of knowledge discovery in databases a skill that has become especially relevant with the explosion in large–scale databases.
Timely, practical, and innovative, this book offers a number of new algorithms and draws on real–world examples including financial and medical applications. It also includes two chapters on loans/credit applications that help identify bad risks and good customers useful for those working with credit scoring and bad debt analysis.
Machine Learning and Statistics is an invaluable resource for researchers involved with artificial intelligence and ML in academia, government, or industry, as well as those working with pattern recognition in statistical departments; for students at the graduate level who seek to expand their horizons; and for anyone who would like to learn more about these cutting–edge methodologies.
The first book to explore the theory, methods, and applications of the relationship between machine learning and statistics.
Here is the first book to address the growing demand for applications that integrate machine learning and statistics. The only text to focus on the interface between the two disciplines, this volume explores a mutually beneficial relationship that is fast becoming recognized by scientists, engineers, and researchers in data analysis and intelligent systems worldwide.
The book shows that machine learning shares several areas of common research with statistics, most notably classification, prediction, and control. It demonstrates that statistical and probability methodologies can be applied in developing different machine learning algorithms and that these algorithms can be used by statisticians to perform classification and forecasting tasks.
Offering an accessible treatment geared to a diverse audience, this collection of contributions from an international group of leading researchers in both fields
Describes numerous applications drawn from real–world projects in finance, investing, medicine, and other areas
- Develops new research topics such as probability trees and prediction
- Outlines new algorithms
- Covers the latest developments in knowledge discovery in systems, addressing the worldwide exponential growth in databases
- Examines the prospects of learning rules in an "Occam′s razor" fashion, using the simplest representation to learn by analogy
- Features over eighty illustrations as well as many references for further reading
