Title :
Discernibility-based algorithm for discretizing continuous variables of Credal network
Author :
Qu, Ying ; Li, Qing-Heng ; Jia, Jian
Author_Institution :
Sch. of Econ. & Manage., Hebei Univ. of Sci. & Technol., Shijiazhuang, China
Abstract :
Having analyzed that the discretization algorithm of rough set and Boolean reasoning approach didn´t work well in Bayesian networks, a new algorithm for discretizing continuous variables is put forward to distinguish two samples by the value of candidate cuts while not by the intervals determined by two candidate cuts. The application case indicates that the improved algorithm can reduce preferably the space complexity and time complexity of the discretization. It is effective on discretizing continuous variables of Credal network.
Keywords :
computational complexity; inference mechanisms; rough set theory; Boolean reasoning approach; Credal network; discernibility-based algorithm; discretizing continuous variables; rough set theory; space complexity; time complexity; Algorithm design and analysis; Bayesian methods; Conference management; Cybernetics; Electronic mail; Gaussian distribution; Machine learning; Machine learning algorithms; Set theory; Technology management; Candidate cuts; Credal network; Rough Set theory; Variable discretization;
Conference_Titel :
Machine Learning and Cybernetics, 2009 International Conference on
Conference_Location :
Baoding
Print_ISBN :
978-1-4244-3702-3
Electronic_ISBN :
978-1-4244-3703-0
DOI :
10.1109/ICMLC.2009.5212635
