A charged system search approach for data clustering

An Enhanced Version of Cat Swarm Optimization Algorithm for Cluster Analysis

Clustering is an unsupervised machine learning technique that optimally organizes the data objects in a group of clusters. In present work, a meta-heuristic algorithm based on cat intelligence is adopted for optimizing clustering problems. Further, to make the cat swarm algorithm (CSO) more robust for partitional clustering, some modifications are incorporated in it. These modifications include an improved solution search equation for balancing global and local searches, accelerated velocity equation for addressing diversity, especially in tracing mode. Furthermore, a neighborhood-based search strategy is introduced to handle the local optima and premature convergence problems. The performance of enhanced cat swarm optimization (ECSO) algorithm is tested on eight real-life datasets and compared with the well-known clustering algorithms. The simulation results confirm that the proposed algorithm attains the optimal results than other clustering algorithms.

Cellular Automata Based Model for E-Healthcare Data Analysis

E-healthcare is warm area of research and a number of algorithms have been applied to classify healthcare data. In the healthcare field, a large amount of clinical data is generated through MRI, CT scans, and other diagnostic tools. Healthcare analytics are used to analyze the clinical data of patient records, disease diagnosis, cost, hospital management, etc. Analytical techniques and data visualization are used to get the real time information. Further, this information can be used for decision making. Also, this information is useful for the better treatment of patients. In this work, an improved big bang-big crunch (BB-BC) based clustering algorithm is applied to analyze healthcare data. Cluster analysis is an important task in the field of data analysis and can be used to understand the organization of data. In this work, two healthcare datasets, CMC and cancer, are used and the proposed algorithm obtains better results when compared to MEBB-BC, BB-BC, GA, PSO and K-means algorithms. The performance of the improved BB-BC algorithm is also examined against benchmark clustering datasets. The simulation results showed that proposed algorithm improves the clustering results significantly when compared to other algorithms.

A Review on Artificial Bee Colony Algorithms and Their Applications to Data Clustering

Data clustering is an important data mining technique being widely used in numerous applications. It is a method of creating groups (clusters) of objects, in such a way that objects in one cluster are very similar and objects in different clusters are quite distinct, i.e. intra-cluster distance is minimized and inter-cluster distance is maximized. However, the popular conventional clustering algorithms have shortcomings such as dependency on center initialization, slow convergence rate, local optima trap, etc. Artificial Bee Colony (ABC) algorithm is one of the popular swarm based algorithm inspired by intelligent foraging behaviour of honeybees that helps to minimize these shortcomings. In the past, many swarm intelligence based techniques for clustering were introduced and proved their performance. This paper provides a literature survey on ABC, its variants and its applications in data clustering.