J. V. R. Murthy

Particle swarm optimized multiple regression linear model for data classification

This paper presents a new data classification method based on particle swarm optimization (PSO) techniques. The paper discusses the building of a classifier model based on multiple regression linear approach. The coefficients of multiple regression linear models (MRLMs) are estimated using least square estimation technique and PSO techniques for percentage of correct classification performance comparisons. The mathematical models are developed for many real world datasets collected from UCI machine repository. The mathematical models give the user an insight into how the attributes are interrelated to predict the class membership. The proposed approach is illustrated on many real data sets for classification purposes. The comparison results on the illustrative examples show that the PSO based approach is superior to traditional least square approach in classifying multi-class data sets.

An Efficient Hybrid Algorithm for Data Clustering Using Improved Genetic Algorithm and Nelder Mead Simplex Search

Data clustering is a process of putting similar data into groups. A clustering algorithm partitions a data set into several groups such that the similarity within a group is larger than among groups. This paper presents data clustering using improved genetic algorithm (IGA) and the popular Nelder-Mead(NM) Simplex search . To improve the accuracy of data clustering, an improved GA (IGA) is used. The performance of IGA is established with many benchmark test functions optimization. To accelerate the clustering process further more a hybrid algorithm based on improved GA and Nelder-Mead simplex search(NM) is suggested for clustering and is tested on 7 datasets and its performance is compared with above two algorithms and the traditional K-means algorithm.

Cluster Analysis on Different Data Sets Using K-Modes and K-Prototype Algorithms

The k-means algorithm is well-known for its efficiency in clustering large data sets and it is restricted to the numerical data types. But the real world is a mixture of various data typed objects. In this paper we implemented algorithms which extend the k-means algorithm to categorical domains by using Modified k-modes algorithm and domains with mixed categorical and numerical values by using k-prototypes algorithm. The Modified k-modes algorithm will replace the means with the modes of the clusters by following three measures like “using a simple matching dissimilarity measure for categorical data”, “replacing means of clusters by modes” and “using a frequency-based method to find the modes of a problem used by the k-means algorithm”. The other algorithm used in this paper is the k-prototypes algorithm which is implemented by integrating the Incremental k-means and the Modified k-modes partition clustering algorithms. All these algorithms reduce the cost function value.

An Approach to Measure Security of Cloud Hosted Application

Recent time has seen rampant proliferation in the cloud services. However cloud services characteristics has also increased opportunities for attacks on data and applications hosted in the cloud. Unsecure practices at technical or/and at operational level by cloud service providers (CSP), dependency on web based service delivery and cloud technology related vulnerabilities could lead to application and data compromise. These threat scenarios require cloud customer to look for more transparency and controls. SLAs and contracts do not provide technical and measurable method to find the security control status of cloud hosted application/data. In this paper,we propose an approach and system for security measurement, which can help cloud customers to have realistic view of their application/data security posture in the cloud environment.

A survey of cross-domain text categorization techniques

Text Mining is important, emerging, research area, because plenty of text resources growing rapidly through the internet and digital world. In the text data analysis text categorization is one of the vital techniques. Traditional text categorization methods are not able to handle well with learning across different domains. Cross-domain classification is more challenging problem than single domain classification problem. In this paper survey of cross-domain text categorization techniques have been presented.

A Comparative Analysis of Unsupervised K-Means, PSO and Self-Organizing PSO for Image Clustering

This paper presents a comparative analysis of three algorithms namely K-means, Particle swarm Optimization (PSO) and Self-Organizing PSO (SOPSO) for image clustering problems. The traditional K-means algorithm found to be trapped in local minima. However, PSO and SOPSO overcome the problem of local minima and provide better results. In this work gbest model is used in PSO and both West and gbest models are used in SOPSO based on self-Organizing rules. It is shown that PSO and SOPSO produce better results compared to K-means with respect to the quantization error, inter- and intra-cluster distances.

A scalable genetic programming multi-class ensemble classifier

In this paper we propose a simple scalable genetic programming multi-class ensemble classifier of higher accuracy. A formula is derived to obtain the maximum number of nodes permitted in a GP classifier. A wrapper approach for feature selection mechanism based on GP classifier is adopted in our work.

Homogeneity Separateness: A New Validity Measure for Clustering Problems

Several validity indices have been designed to evaluate solutions obtained by clustering algorithms. Traditional indices are generally designed to evaluate center-based clustering, where clusters are assumed to be of globular shapes with defined centers or representatives. Therefore they are not suitable to evaluate clusters of arbitrary shapes, sizes and densities, where clusters have no defined centers or representatives. In this work, HS (Homogeneity Separateness) validity measure based on a different shape is proposed. It is suitable for clusters of any shapes, sizes and/or of different densities. The main concepts of the proposed measure are explained and experimental results on both synthetic and real life data set that support the proposed measure are given.

Performance of Teaching Learning Based Optimization Algorithm with Various Teaching Factor Values for Solving Optimization Problems

Teaching Learning Based Optimization (TLBO) is being used as a new, reliable, accurate and robust optimization technique scheme for global optimization over continuous spaces. This paper presents an effect of variation of a teaching factor TF in traditional TLBO algorithm and then proposed a value for teaching factor TF. The traditional TLBO algorithm with new teaching factor TF value has been tested on several benchmark functions and shown to be statistically significantly better than other teaching factor values for performance measures in terms of faster convergence behavior.

A New Approach of Integrating PSO & Improved GA for Clustering with Parallel and Transitional Technique

This paper studies the applicability of hybridization of Improved GA (IGA) and PSO techniques to data clustering problem. A new way of integrating IGA and PSO is explored in the paper. In one approach, a parallel IGA and PSO developed and in other, a transitional approach of alternate IGA and PSO technique followed Simulations for number of data sets show that the proposed integrated approach provides better clustering performance.