Ranking and Empirical Minimization of U-statistics
Clémençon, Stéphan ; Lugosi, Gábor ; Vayatis, Nicolas
Ann. Statist., Tome 36 (2008) no. 1, p. 844-874 / Harvested from Project Euclid
The problem of ranking/ordering instances, instead of simply classifying them, has recently gained much attention in machine learning. In this paper we formulate the ranking problem in a rigorous statistical framework. The goal is to learn a ranking rule for deciding, among two instances, which one is “better,” with minimum ranking risk. Since the natural estimates of the risk are of the form of a U-statistic, results of the theory of U-processes are required for investigating the consistency of empirical risk minimizers. We establish, in particular, a tail inequality for degenerate U-processes, and apply it for showing that fast rates of convergence may be achieved under specific noise assumptions, just like in classification. Convex risk minimization methods are also studied.
Publié le : 2008-04-15
Classification:  Statistical learning,  theory of classification,  VC classes,  fast rates,  convex risk minimization,  moment inequalities,  U-processes,  68Q32,  60E15,  60C05,  60G25
@article{1205420521,
     author = {Cl\'emen\c con, St\'ephan and Lugosi, G\'abor and Vayatis, Nicolas},
     title = {Ranking and Empirical Minimization of U-statistics},
     journal = {Ann. Statist.},
     volume = {36},
     number = {1},
     year = {2008},
     pages = { 844-874},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1205420521}
}
Clémençon, Stéphan; Lugosi, Gábor; Vayatis, Nicolas. Ranking and Empirical Minimization of U-statistics. Ann. Statist., Tome 36 (2008) no. 1, pp.  844-874. http://gdmltest.u-ga.fr/item/1205420521/