Margin-based Ranking and an Equivalence between AdaBoost and RankBoost

By Cynthia Rudin, Robert E. Schapire. 2009

109 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Direct Learning to Rank and Rerank (2018, methodology): “…guarantee on ranking methods is obtained by directly optimizing the rank statistic of interest (as shown by theoretical bounds of Clemenc¸on and Vayatis, 2008; Rudin and Schapire, 2009, for instance) hence our choice of methodology – mixed-integer programming (MIP) – for reranking in this work.”
All Models are Wrong but many are Useful: Variable Importance for Black-Box, Proprietary, or Misspecified Prediction Models, using Model Class Reliance (2018, background): “The bounds in Theorem 6 naturally account for potential overfitting without an explicit limit on model class complexity, such as a covering number (Bousquet et al., 2004; Rudin and Schapire, 2009).”
Machine learning with operational costs (2011, methodology): “The main ideas of Maurey’s Lemma are used in many machine learning papers in various contexts (e.g., Koltchinskii and Panchenko, 2005; Schapire et al., 1998; Rudin and Schapire, 2009).”
Surrogate regret bounds for bipartite ranking via strongly proper losses (2012, methodology): “Nevertheless, it has often been observed that algorithms such as AdaBoost, logistic regression, and in some cases even SVMs, which minimize the exponential, logistic, and hinge losses respectively in the standard (non-pairwise) setting, also yield good bipartite ranking performance [11, 20, 24].”
On Equivalence Relationships Between Classification and Ranking Algorithms (2011, methodology): “The corresponding proof within Rudin and Schapire (2009) used fourmain steps, which we follow also in our proof: 1) characterizing the conditions to be at a minimizer of the classification objective function, 2) using those conditions to develop a “skew” condition on the classes, 3) characterizing…”
One-Pass AUC Optimization (2013, unknown): “ et al., 1998; Cortes and Mohri, 2004; Liu et al., 2009; Flach et al., 2011). Many algorithms have been developed to optimize AUC based on surrogate losses (Herschtal and Raskutti,2004;Joachims, 2006;Rudin and Schapire, 2009; Kotlowski et al., 2011; Zhao et al., 2011). In this work, we focus on AUC ”
The P-Norm Push: A Simple Convex Ranking Algorithm that Concentrates at the Top of the List (2009, methodology): “On the other hand,f2 misses the top two examples which have scores -1/14 and -2/14, however, the 10 middle examples are correctly ranked.f2 has a larger AUC thanf1, but f1 is better at the top of the list.”