Performance Enhancement of K Nearest Neighbor Classification Algorithm Using 8-Bin Hashing and Feature Weighting
Akhil Rane, Nitesh Naik, J. A. Laxminarayana · 2014
The K-Nearest Neighbor (K-NN) algorithm is an instance based learning method that has been widely used in many pattern classification tasks due to its simplicity, effectiveness and robustness. Standard K-NN fails to work satisfactorily due to many limitations. This paper deals with improvements over two major limitations. First, to improve the efficiency of classification by eliminating those instances which are too far away from query instance using 8-bin hashing technique thus reducing the computational time. Second, to improve accuracy of classification by weighting the features using positive instances based feature weighting algorithm. Finally, this paper gives validation of improvement theories through experimental results.