This important text and reference for researchers and students in machine learning, game theory, statistics and information theory offers a comprehensive treatment of the problem of predicting individual sequences. Unlike standard statistical approaches to forecasting, prediction of individual sequences does not impose any probabilistic assumption on the data-generating mechanism. Yet, prediction algorithms can be constructed that work well for all possible sequences, in the sense that their performance is always nearly as good as the best forecasting strategy in a given reference class. The central theme is the model of prediction using expert advice, a general framework within which many related problems can be cast and discussed. Repeated game playing, adaptive data compression, sequential investment in the stock market, sequential pattern analysis, and several other problems are viewed as instances of the experts' framework and analyzed from a common nonstochastic standpoint that often reveals new and intriguing connections.
1. Introduction; 2. Prediction with expert advice; 3. Tight bounds for specific losses; 4. Randomized prediction; 5. Efficient forecasters for large classes of experts; 6. Prediction with limited feedback; 7. Prediction and playing games; 8. Absolute loss; 9. Logarithmic loss; 10. Sequential investment; 11. Linear pattern recognition; 12. Linear classification; 13. Appendix.
"Each chapter contains about two dozen inspiring problems, and the book refers to more than 300 up-to-date sources.... The book is addressed to graduate students and researchers in the fields of engineering and information, computer sciences, and data analysis; it presents both theoretical inference and practical hands-on usage of modern prediction techniques."
Stan Lipovektsy, GfK Custom Research North AmericaTechnometrics
"This book is a comprehensive treatment of current results on predicting using expert advice."
David S. Leslie, Mathematical Reviews