Variable Neighborhood Search Based Human Learning Optimization Algorithm for Secure Data Analysis and Computing

The data analysis in medical field is a very crucial task in order to gain insights from a large collection of data. The analysis of data comprises of several well-defined processes like data collection, data preprocessing and exploratory data analysis techniques to compute undiscovered patterns and trends. Several statistical and machine learning algorithms are utilized to discover meaningful patterns and information. Clustering is an unsupervised machine learning technique that aims at discovering patterns in data to form clusters of similar data objects. In this work, an effective diagnostic model is being proposed based upon human learning optimization (HLO) algorithm. Since the HLO algorithm have shortcomings like trade-off issues and local optima, an improved search mechanism and variable neighborhood strategy is proposed in HLO to deal with them. The proposed technique is evaluated using real-world benchmark clustering datasets. The results are assessed using well-known performance parameters like accuracy, detection rate and f-measure and compared with several existing algorithms while detection rate is improved by 4%. The outcomes obtained reflects the supremacy of the proposed technique in effective cluster analysis.