A Probability Model for Combining Ranks

Mixed Group Ranks is a parametric method for combining rank based classiers that is eective for many-class problems. Its parametric structure combines qualities of voting methods with best rank approaches. In [1] the parameters of MGR were estimated using a logistic loss function. In this paper we describe how MGR can be cast as a probability model. In particular we show that using an exponential probability model, an algorithm for ecient maximum likelihood estimation of its parameters can be devised. While casting MGR as an exponential probability model oers provable asymptotic properties (consistency), the interpretability of probabilities allows for exiblity and natural integration of MGR mixture models. 1 MGR as a Score Function Many rank combination approaches can be cast as the problem of assigning scores to classes based on the ranks they receive from multiple consituent classiers. Once assigned, then classes can be ordered based on their scores, generating a combined rank...

• Type: Conference
• Year: 2005
• Where: MCS
• Authors: Ofer Melnik, Yehuda Vardi, Cun-Hui Zhang
• Keywords: Exponential Probability Model, MCS 2005, Pattern Recognition, Probability Model, Score Functions