A non-parametric semi-supervised discretization method |
| |
Authors: | Alexis Bondu Marc Boull�� Vincent Lemaire |
| |
Affiliation: | (1) Computer Science Department, College of Staten Island of CUNY, 2800 Victory Blvd, Staten Island, NY 10314, USA;(2) Center for Automated Learning and Discovery, Carnegie Mellon University, 500 Forbes Avenue, Pittsburgh, PA 15213, USA;(3) Department of Computer Science, Rutgers University, 110 Frelinghuysen Road, Piscataway, NJ 08854-8019, USA |
| |
Abstract: | Semi-supervised classification methods aim to exploit labeled and unlabeled examples to train a predictive model. Most of
these approaches make assumptions on the distribution of classes. This article first proposes a new semi-supervised discretization
method, which adopts very low informative prior on data. This method discretizes the numerical domain of a continuous input
variable, while keeping the information relative to the prediction of classes. Then, an in-depth comparison of this semi-supervised
method with the original supervised MODL approach is presented. We demonstrate that the semi-supervised approach is asymptotically
equivalent to the supervised approach, improved with a post-optimization of the intervals bounds location. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|