K. Rajeswari

Improvement in k-Means Clustering Algorithm Using Data Clustering

The set of objects having same characteristics are organized in groups and clusters of these objects reformed known as Data Clustering. It is an unsupervised learning technique for classification of data. K-means algorithm is widely used and famous algorithm for analysis of clusters. In this algorithm, n number of data points are divided into k clusters based on some similarity measurement criterion. K-Means Algorithm has fast speed and thus is used commonly clustering algorithm. Vector quantization, cluster analysis, feature learning are some of the application of K-Means. However results generated using this algorithm are mainly dependant on choosing initial cluster centroids. The main short come of this algorithm is to provide appropriate number of clusters. Provision of number of clusters before applying the algorithm is highly impractical and requires deep knowledge of clustering field. In this project, we are going to propose an algorithm for improvement in the initializing the centroids for K-Means algorithm. We are going to work on numerical data sets along with the categorical datasets with the n dimensions. For similarity measurement we are going to consider the Manhattan distance,Dice distance and cosine distance. The result of this proposed algorithm will be compared with the original K-Means. Also the quality and complexity of the proposed algorithm will be checked with the existing algorithm.

Analysis of Different Classifiers for Medical Dataset using Various Measures

The process of extracting information from a dataset and transforming it into an understandable structure for further use is called as data mining. A number of important techniques such as preprocessing, classification, clustering are performed in data mining using WEKA tool. In medical diagnoses the role of data mining approaches is being increased. Particularly Classification algorithms are very helpful in classifying the data, which is important for decision making process for medical practitioners. To increase the accuracy in the short time ensemble is used. The ensemble is formed by combination of two or more classifiers. For experimentation of ensembles, different types of base classifiers such as Bagging and Adaboost in combination with classifiers and classifiers such as C4.5, J48, and AD tree are used in the medical data set. The experiment is carried out in the WEKA tool on the UCI machine repository. Experimental results for ensemble with bagging classifier shows good accuracy for FT Tree in less time. Also arrthmia dataset shows the highest average accuracy.

Implementation of an efficient web crawler to search medicinal plants and relevant diseases

In Indian Ayurvedic system, medicinal plant is the important factor for different therapeutic uses. The medicinal plants are distributed over India. Hence collection of correct information of medicinal plants is needed for various researches. The huge amount of information related to medicinal plants domain is available on the Internet. Focused web crawler is useful for collecting such close domain information from the internet. In focused web crawler, to extract the relevant web page classification method is used. The classification algorithm classifies the web pages as relevant or not for a given query. In this paper, proposed method of an efficient focused web crawler is used to search the web pages for a medicinal plant domain. Naive Bayes classifier is used for classification of web pages. The proposed focused web crawler uses a manual thesaurus of medicinal plant information for query expansion.

GPU accelerated Chemical Text mining for relationship identification between chemical entities in heterogeneous environment

Chemical Text mining techniques have been extensively used to reveal interesting patterns and relationships between proteins, genes, disease and drug from biomedical and chemical literature. Various chemical Text mining tools were proposed for mining chemical data from various chemical databases like PubMed, Drug-Bank, etc. The exponential increase in the generation and collection of data has led to the new era of information extraction and data analysis. Mining the huge and massive literature data on conventional general purpose systems can be a tedious task which is unable to meet high computational requirements of text mining. To accelerate this process parallel Text mining operations can be performed using Graphical Processing Unit (GPU). This paper provides the literature on various Chemical text mining tools and techniques. This paper also presents a scalable framework for high performance GPU accelerated Chemical Text mining for relationship identification between chemical entities on heterogeneous platform providing a method of balancing the workload between CPUs and GPUs for Maximum utilization of diverse commodity hardware.

Causal Association Mining for Detection of Adverse Drug Reactions

Adverse drug reactions (ADRs) are the harmful reactions of the drugs caused to humans due to allergies, overdose, chemical reactions between two chemicals in the medicines, etc. To reduce these reactions is a very important task so as to save lives of the patients as ADRs are a serious topic nowadays [4]. Detecting such harmful effects as early as possible is a very important to prevent harmful consequences. Therefore, mining causal relationships between the drug related events is essential. A method for detecting the potential relationship between drug and ICD is done using causal association rules suitable for the frequent events[1]. Sometimes an infrequent nature of the drugs can cause tremendous harm especially in case of Type-2 diabetes. A new interestingness measure called as exclusive causal leverage can be used based on fuzzy Recognition Primed Decision model (RPD) [3]. On the basis of this measure the relationship between the drug and associated drug reactions can be mined.

An ensemble approach for cancerious dataset analysis using feature selection

Feature selection (FS) is an important technique in data mining to remove noise, irrelevant and redundant data. The paper introduces the ensemble approach using FS and without using FS tested on a standard medical dataset in order to compare the accuracy and time of both. This system uses best first search FS algorithm to reduce the noise in the dataset. The ensemble technique is a combination of two or more classifiers i.e. meta classifiers and classifiers. Bagging, Boosting and Adaboost are meta classifiers. In the proposed work Bagging and Adaboost ensembles are used, but the main focus is on Bagging Ensembles as it has been proven best compared to Adaboost and Boosting ensembles [1]. This paper concludes that better results can be achieved by applying FS on ensembles.

Artificial Intelligence (AI) Techniques Applied for the Development of a Clinical Decision Support System (CDSS) for Diagnosing Ischemic Heart Disease (IHD)

Heart Disease (IHD) is difficult to diagnose since most of the symptoms and clinical presentations are similar to other diseases. It is a very common, harmful disease, which is identified mostly during the mortality of an individual. The objective is to build a clinical decision support system, which will diagnose the presence of IHD with an integrated automated classifier using Artificial Intelligence (AI) techniques. A retrospective data set that included 800 clinical cases was taken for the work. A total of 88 sets were discarded during pre-processing. Tests were run on 712 cases using the Weka classifiers available in Weka 3.7.0. Out of 113 classifiers, 16 were identified to be the best based on the following parameters: sensitivity, specificity, accuracy, F- measure, kappa statistic, correctly classified cases, time taken to run the model, and the Receiver Operating Characteristic (ROC) curve. The diagnoses made by the Clinical Decision Support System (CDSS) were compared with those made by physicians during patient consultations. The KSTAR algorithm showed the best diagnoses with the highest accuracy 97.32%, sensitivity 98%, specificity 97% kappa 0.95, and ROC 0.995. The authors thus conclude that a CDSS can be developed to assist expert physicians in separating the positive and the negative cases of heart disease.