Abstract | ||
---|---|---|
Naïve Bayesian classifier is one of the most effective and efficient classification algorithms. The elegant simplicity and apparent accuracy of naive Bayes (NB) even when the independence assumption is violated, fosters the on-going interest in the model. Rough Sets Theory has been used for different tasks in knowledge discovery and successfully applied in many real-life problems. In this study we make use of rough sets ability, in discovering attributes dependencies, to overcome the NB un-practical assumption. We propose a new algorithm called Rough-Naive Bayes (RNB) that is expected to outperform other current NB variants. RNB is based on adjusting attributes' weights based on their dependencies and contribution to the final decision. Experimental results show that RNB can achieve better performance than NB classifier. |
Year | DOI | Venue |
---|---|---|
2010 | 10.1007/978-3-642-16248-0_23 | RSKT |
Keywords | Field | DocType |
data classification,rough sets theory,apparent accuracy,current nb variant,rough set,nb un-practical assumption,nb classifier,better performance,attributes dependency,naive bayes,independence assumption,rough-naive bayes,classification,bayesian classifier,rough set theory | Naive Bayes classifier,Pattern recognition,Computer science,Rough set,Artificial intelligence,Data classification,Classifier (linguistics),Statistical classification,Statistical assumption,Bayes classifier,Machine learning,Bayes' theorem | Conference |
Volume | ISSN | ISBN |
6401 | 0302-9743 | 3-642-16247-9 |
Citations | PageRank | References |
2 | 0.37 | 18 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Khadija Al-Aidaroos | 1 | 2 | 0.37 |
Azuraliza Abu Bakar | 2 | 157 | 30.29 |
Zalinda Othman | 3 | 146 | 7.63 |