Abstract | ||
---|---|---|
A new approach for feature selection is presented in this paper. The proposed approach uses the Harmony Search with a novel fitness function to eliminate noisy and irrelevant features. Harmony vectors contain real weights which refer to feature space. The best and significant features are selected according to a threshold. Fitness function of Harmony Search is based on the Area Under the receiver operating characteristics Curve (AUC). All of the selected features are employed to improve the classification of the k Nearest Neighbor (k-NN) classifier. Experimental results claim that the proposed method is able to improve the classification performance of k-NN algorithm in comparison with the other important methods in realm of feature selection such as BAHSIC, FSS, BSS and MFS. |
Year | DOI | Venue |
---|---|---|
2012 | 10.1145/2393216.2393219 | CCSEIT |
Keywords | Field | DocType |
k-nn algorithm,feature selection,significant feature,irrelevant feature,harmony search,fitness function,novel fitness function,harmony-based feature selection,selected feature,nearest neighbor classification,new approach,classification performance | k-nearest neighbors algorithm,Data mining,Feature vector,Receiver operating characteristic,Pattern recognition,Feature selection,Computer science,Fitness function,Harmony search,Artificial intelligence,Classifier (linguistics),Harmony (color) | Conference |
Citations | PageRank | References |
0 | 0.34 | 13 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ali Adeli | 1 | 4 | 1.47 |
Mehrnoosh Sinaee | 2 | 3 | 1.11 |
Javad Zomorodian | 3 | 0 | 0.34 |
Mehdi Neshat | 4 | 83 | 7.08 |