An Enhanced Multi Attribute Depthness Similarity Estimation Technique to Improve Classification Accuracy
N. Elavarasan, K. Mani · 2017
The classification accuracy states the efficiency of the classification algorithm in high dimensional classification problems. To improve the accuracy of classification, many algorithms have been proposed earlier, but they suffer from poor classification accuracy and more false classification ratio. To overcome this issue, a multi attribute depthness similarity estimation technique to classify the data points towards the number of classes has been proposed in this paper. The proposed method takes the dataset and data points grouped by using labels. This method performs preprocessing by removing the noisy data points. It identifies distinct attributes of each class which has more importance among the data points. Based on identified attribute sets, this method computes multi attribute depthness similarity for each of the data points towards each class. The class which has more depthness value will be assigned as a class to the input data point. The classification accuracy is high compared to other algorithms by employing the Pertinent Dimension Detection and classification algorithm.