A Method of Discovering Tolerance Markov Blanket Based on Completely Dependent Unknown Components
Abstract
A novel tolerance feature subset selection method from incomplete data set, denoted by MaxG-IIAMB, is proposed to pick out the Markov-boundary (MB), the minimal subset of features, of target variable but without making any assumption about the unknown component distribution. The classification experimental results of risk factors observed in a sample of 1841 employees of a Czech car factory demonstrate the practicability and superiority of our method over the classical expectation-maximization (EM) and available case technique (ACA).
Keywords
Feature Subset Selection (FSS); Markov Boundary (MB); Probabilistic Classification; Unknown Component Distribution
Full Text:
PDFRefbacks
- There are currently no refbacks.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.