Abstract:Bayesian network (BN), a simple graphical notation for conditional independence assertions, is promised to represent the probabilistic relationships between diseases and symptoms. Learning the structure of a Bayesian network classifier (BNC) encodes conditional independence assumption between attributes, which may deteriorate the classification performance. One major approach to mitigate the BNC's primary weakness (the attributes independence assumption) is the locally weighted approach. And this type of approach has been proved to achieve good performance for naive Bayes, a BNC with simple structure. However, we do not know whether or how effective it works for improving the performance of the complex BNC. In this paper, we first do a survey on the complex structure models for BNCs and their improvements, then carry out a systematically experimental analysis to investigate the effectiveness of locally weighted method for complex BNCs, e.g., tree-augmented naive Bayes (TAN), averaged one-dependence estimators AODE and hidden naive Bayes (HNB), measured by classification accuracy (ACC) and the area under the ROC curve ranking (AUC). Experiments and comparisons on 36 benchmark data sets collected from University of California, Irvine (UCI) in Weka system demonstrate that locally weighting technologies just slightly outperforms unweighted complex BNCs on ACC and AUC. In other words, although locally weighting could significantly improve the performance of NB (a BNC with simple structure), it could not work well on BNCs with complex structures. This is because the performance improvements of BNCs are attributed to their structures not the locally weighting.

Extracting Credible Dependencies for Averaged One-Dependence Estimator Analysis

Self-Adaptive Attribute Value Weighting for Averaged One-Dependence Estimators.

Averaged Tree-Augmented One-Dependence Estimators

Alleviating the Attribute Conditional Independence and I.I.D. Assumptions of Averaged One-Dependence Estimator by Double Weighting

Alleviating the independence assumptions of averaged one-dependence estimators by model weighting

Attribute Value Weighted Average of One-Dependence Estimators

Attribute Value Weighted Averaged One-Dependence Estimators with Kullback–Leibler Divergence

Model Weighting for One-Dependence Estimators by Measuring the Independence Assumptions

Weighted One-Dependence Forests Classifier.

Instance-based weighting filter for superparent one-dependence estimators

Semi-supervised Weighting for Averaged One-Dependence Estimators

Locally Weighted Learning: How and when Does It Work in Bayesian Networks?

Semi-naive Exploitation of One-Dependence Estimators

Applying functional dependencies to improve the performance and robustness of hidden naive bayes

Learning by Extrapolation from Marginal to Full-Multivariate Probability Distributions: Decreasingly Naive Bayesian Classification

General and Local: Averaged K-Dependence Bayesian Classifiers

Robust Structure Learning of Bayesian Network by Identifying Significant Dependencies

A Novel Approach to Fully Representing the Diversity in Conditional Dependencies for Learning Bayesian Network Classifier

To Select or to Weigh: A Comparative Study of Linear Combination Schemes for SuperParent-One-Dependence Estimators

Identification of Informational and Probabilistic Independence by Adaptive Thresholding

Bagging K-Dependence Bayesian Network Classifiers