Abstract

A radically new approach to statistical modelling, which combines mathematical techniques of Bayesian statistics with the philosophy of the theory of competitive on-line algorithms, has arisen over the last decade in computer science (to a large degree, under the influence of Dawid’s prequential statistics). In this approach, which we call “competitive on-line statistics”, it is not assumed that data are generated by some stochastic mechanism; the bounds derived for the performance of competitive on-line statistical procedures are guaranteed to hold (and not just hold with high probability or on the average). This paper reviews some results in this area; the new material in it includes the proofs for the performance of the Aggregating Algorithm in the problem of linear regression with square loss. Keywords: Bayes’s rule, competitive on-line algorithms, linear regression, prequential statistics, worst-case analysis.

Citations

1714	Schapire R: A decision-theoretic generalization of online learning and an application to boosting - Freund - 1997
1412	An introduction to Kolmogorov complexity and its applications, Second edn - Li, Vitanyi - 1997
556	The weighted majority algorithm - Littlestone, Warmuth - 1994
399	Applied Regression Analysis - DRAPER, SMITH - 1966
325	A universal prior for the integers and estimation by minimum description length. Ann Stat - Rissanen - 1983
316	A Formal Theory of Inductive Inference - Solomonoff - 1964
269	An Information Measure for Classification - Wallace, Boulton - 1968
267	How to use expert advice - Cesa-Bianchi, Freund, et al. - 1997
213	Context-sensitive learning methods for text categorization - Cohen, Singer - 1999
211	Aggregating strategies - Vovk - 1990
196	Exponentiated gradient versus gradient descent for linear predictors - Kivinen, Warmuth - 1997
157	Tracking the best expert - Herbster, Warmuth - 1995
144	Gambling in a rigged casino: the adversarial multiarmed bandit problem - Auer, Cesa-Bianchi, et al. - 1995
144	A dynamic disk spin-down technique for mobile computing - Helmbold, Long, et al. - 1996
131	Statistical Theory: The Prequential Approach - Dawid - 1984
122	Universal portfolios - Cover - 1991
117	Game theory, On-line Prediction and Boosting - Freund, Schapire - 1996
106	Adaptive game playing using multiplicative weights - Freund, Schapire - 1999
86	A game of prediction with expert advice - Vovk - 1998
83	M.K.: Relative loss bounds for on-line density estimation with the exponential family of distributions - Azoury, Warmouth - 2001
76	Using and combining predictors that specialize - Freund, Schapire, et al. - 1997
65	Universal portfolios with side information - Cover, Ordentlich - 1996
65	The Minimum Description Length Principle and Reasoning - Grünwald - 1998
64	Tracking the best disjunction - Auer, Warmuth - 1995
64	Predicting nearly as well as the best pruning of a decision tree - Helmbold, Schapire - 1997
58	Sequential prediction of individual sequences under general loss functions - Haussler, Kivinen, et al. - 1998
51	Tight worst-case loss bounds for predicting with expert advice - HAUSSLER, KIVINEN, et al. - 1995
44	Universal portfolios with and without transaction costs - Blum, Kalai - 1997
41	Asymptotic minimax regret for data compression, gambling and prediction - Xie, Barron
41	A decision-theoretic extension of stochastic complexity and its applications to learning - Yamanishi - 1995
40	Averaging expert predictions - Kivinen, Warmuth
35	Predicting a binary sequence almost as well as the optimal biased coin - Freund - 2003
34	Derandomizing stochastic prediction strategies - Vovk - 1999
32	Probability forecasting - Dawid - 1986
31	On-line learning and the metrical task system problem - Blum, Burch - 1997
31	Learning Probabilistic Prediction Functions - DeSantis, Markowski, et al. - 1988
29	Worstcase quadratic loss bounds for on-line prediction of linear functions by gradient descent - Cesa-Bianchi, Long, et al. - 1996
26	Stochastic Complexity" (with discussion - Rissanen - 1987
23	An efficient extension to mixture techniques for prediction and decision trees - Pereira, Singer - 1999
22	Machine-learning applications of algorithmic randomness - Vovk, Gammerman, et al. - 1999
19	Universal portfolio selection - Vovk, Watkins - 1998
18	Fisherian inference in likelihood and prequential frames of reference (with - Dawid - 1991
17	Prediction in the worst case - Foster - 1991
17	Tracking the best regressor - Herbster, Warmuth - 1998
16	A logic of probability, with application to the foundations of statistics - VOVK - 1993
16	Competitive on-line linear regression - Vovk - 1998
15	Information theoretic asymptotics of bayes’ methods - Clarke, Barron - 1989
13	On a randomness criterion - Vovk - 1987
12	Estimation and inference by compact coding (with discuss). J.R.Statist Soc - WALLACE, FREEMAN - 1987
11	Universal forecasting algorithms - VOVK - 1992