High Dimensional Data Clustering Using Forward Feature Selection Framework for Medical Data

Abstract

AbstractThe advent of the “curse of dimensionality” issue while dealing with medical data as a result of large reduced datasets weakens the power of learning algorithms and requires expensive memory and processing expenses. Feature selection helps improve the performance of machine learning algorithms by decreasing the time it takes to create a learning model and enhancing the accuracy of the learning process. The High Dimensional Forward Feature Selection Clustering (HDFFSC) technique is proposed in this paper, which is a simple yet effective parallel processing method based on Map Reduce. Even though numerous algorithms have been created, they still fall short when dealing with high dimensional data in the medical area. To address the challenges with high dimensional data in the medical area, this paper also provides a Forward Feature Selection (FFS) technique with Artificial Bee Colony (ABC) optimization. Between and between clusters, this architecture enabled both global and local search capabilities. In comparison to standard clustering methods, it also increases clustering performance over chosen UCI medical datasets.KeywordsArtificial Bee Colony (ABC)Forward feature selection (FFS)ClusteringHigh dimensional dataOptimization algorithms