A Novel Mixed-Attribute Fusion-Based Naive Bayesian Classifier

The Naive Bayesian classifier (NBC) is a well-known classification model that has simple structure, low training complexity, excellent scalability, and good performances. However, the NBC two key limitations: (1) it built upon strong assumption condition attributes are independent, which often does not hold in real-life, (2) handle continuous well. To overcome these limitations, this paper presents novel approach for construction, called mixed-attribute fusion-based (MAF-NBC). It alleviates aforementioned limitations by relying on fusion mechanism with an improved autoencoder neural network construction. MAF-NBC transforms original mixed of data set into series encoded maximum independence as pre-processing step. guarantee generation useful attributes, efficient objective function designed to optimize weights considering both encoding error attribute’s dependence. A persuasive experiments was conducted validate feasibility, rationality, effectiveness approach. Results demonstrate superior performance than eight state-of-the-art algorithms, namely discretization-based (Dis-NBC), flexible naive Bayes (FNB), tree-augmented (TAN) Bayes, averaged one-dependent estimator (AODE), hidden (HNB), deep feature weighting (DFW-NBC), correlation-based filter (CFW-NBC), independent component analysis-based (ICA-NBC).