Machine Learning in Non-Stationary Environments ─ Introduction to Covariate Shift Adaptation

As the power of computing has grown over the past few decades, the field of machinelearning has advanced rapidly in both theory and practice. Machine learning methods are usuallybased on the assumption that the data generation mechanism does not change over time. Yet real-worldapplications of machine learning, including image recognition, natural language processing, speechrecognition, robot control, and bioinformatics, often violate this common assumption. Dealing withnon-stationarity is one of modern machine learning's greatest challenges. This book focuses on aspecific non-stationary environment known as covariate shift, in which the distributions of inputs(queries) change but the conditional distribution of outputs (answers) is unchanged, and presentsmachine learning theory, algorithms, and applications to overcome this variety of non-stationarity.After reviewing the state-of-the-art research in the field, the authors discuss topics that includelearning under covariate shift, model selection, importance estimation, and active learning. Theydescribe such real world applications of covariate shift adaption as brain-computer interface,speaker identification, and age prediction from facial images. With this book, they aim to encouragefuture research in machine learning, statistics, and engineering that strives to create trulyautonomous learning machines able to learn under non-stationarity.

Masashi Sugiyama is Associate Professor in the Department of Computer Science at Tokyo Institute of Technology. Motoaki Kawanabe is a Postdoctoral Researcher in Intelligent Data Analysis at the Fraunhofer FIRST Institute, Berlin. In October 2011, he moved to Advanced Telecommunications Research Institute International (ATR) in Kyoto, Japan.