An Improved Multi-label k-Nearest Neighbour Algorithm with Prototype Selection using DENCLUE

Prototype Selection (PS) techniques can be implemented very efficiently to allow the nearest neighbor classification to be faster by using a simple method to preserve the most prominent data for the Multi-Label k-Nearest Neighbour (ML-kNN) training. The most often used algorithm for classifying multi-labeled data is ML-kNN data which is adopted through the use of the well-known kNN algorithm. The use of Prototype Selection often minimizes the accuracy, and hence the performance of the algorithm. To solve the problem of reducing the accuracy of the ML-kNN, we proposed a novel solution that reduces data by using the well-known clustering algorithm-DENCLUE. Data that belongs to one of the clusters is considered prominent, while data that does not belong to any of the clusters is considered noisy and will not be included in the training process. In terms of noisy data reduction, the findings demonstrate a significant improvement over the ML-kNN.