H. Hannah Inbarani

Supervised hybrid feature selection based on PSO and rough sets for medical diagnosis

Medical datasets are often classified by a large number of disease measurements and a relatively small number of patient records. All these measurements (features) are not important or irrelevant/noisy. These features may be especially harmful in the case of relatively small training sets, where this irrelevancy and redundancy is harder to evaluate. On the other hand, this extreme number of features carries the problem of memory usage in order to represent the dataset. Feature Selection (FS) is a solution that involves finding a subset of prominent features to improve predictive accuracy and to remove the redundant features. Thus, the learning model receives a concise structure without forfeiting the predictive accuracy built by using only the selected prominent features. Therefore, nowadays, FS is an essential part of knowledge discovery. In this study, new supervised feature selection methods based on hybridization of Particle Swarm Optimization (PSO), PSO based Relative Reduct (PSO-RR) and PSO based Quick Reduct (PSO-QR) are presented for the diseases diagnosis. The experimental result on several standard medical datasets proves the efficiency of the proposed technique as well as enhancements over the existing feature selection techniques.

Rough Set Based Feature Selection for Web Usage Mining

Web usage mining exploits data mining techniques to discover valuable information from navigation behavior of World Wide Web (WWW) users. The required information is captured by Web servers and stored in Web usage data logs. The first phase of Web usage mining is the pre processing phase. In the preprocessing phase, first, relevant information is filtered from the logs. Data preprocessing is a critical step in Web usage mining. The results of data preprocessing is relevant to the next steps, such as transaction identification, path analysis, association rule mining, sequential pattern mining, and so forth. Feature selection is a preprocessing step in data mining, and it is very effective in reducing dimensions, reducing the irrelevant data, increasing the learning accuracy and improving comprehensiveness. This paper proposes a novel approach for feature selection based on rough set theory for Web usage mining.

Bijective soft set based classification of medical data

Classification is one of the main issues in Data Mining Research fields. The classification difficulties in medical area frequently classify medical dataset based on the result of medical diagnosis or description of medical treatment by the medical specialist. The Extensive amounts of information and data warehouse in medical databases need the development of specialized tools for storing, retrieving, investigation, and effectiveness usage of stored knowledge and data. Intelligent methods such as neural networks, fuzzy sets, decision trees, and expert systems are, slowly but steadily, applied in the medical fields. Recently, Bijective soft set theory has been proposed as a new intelligent technique for the discovery of data dependencies, data reduction, classification and rule generation from databases. In this paper, we present a novel approach based on Bijective soft sets for the generation of classification rules from the data set. Investigational results from applying the Bijective soft set analysis to the set of data samples are given and evaluated. In addition, the generated rules are also compared to the well-known decision tree classifier algorithm and Naïve bayes. The learning illustrates that the theory of Bijective soft set seems to be a valuable tool for inductive learning and provides a valuable support for building expert systems.

Hybrid Tolerance Rough Set-Firefly based supervised feature selection for MRI brain tumor image classification

Brain tumor is one of the most harmful diseases, and has affected majority of people including children in the world. The probability of survival can be enhanced if the tumor is detected at its premature stage. The intention of feature selection approach is to select a small subset of features which minimizes redundancy and maximizes relevance to the target such as the class labels in classification. Thus, the machine learning model receives a brief organization with high predictive accuracy using the selected prominent features. Therefore, currently, feature selection plays a significant role in machine learning and knowledge discovery. A novel hybrid supervised feature selection algorithm, called TRSFFQR (Tolerance Rough Set Firefly based Quick Reduct), is developed and applied for MRI brain images. The hybrid intelligent system aims to exploit the benefits of the basic models and at the same time, moderate their limitations. Different categories of features are extracted from the segmented MRI images, i.e., shape, intensity and texture based features. The features extracted from brain tumor Images are real values. Hence Tolerance Rough set is applied in this work. In this study, a hybridization of two techniques, Tolerance Rough Set (TRS) and Firefly Algorithm (FA) are used to select the imperative features of brain tumor. Performance of TRSFFQR is compared with Artificial Bee Colony (ABC), Cuckoo Search Algorithm (CSA), Supervised Tolerance Rough Set-PSO based Relative Reduct (STRSPSO-RR) and Supervised Tolerance Rough Set-PSO based Quick Reduct (STRSPSO-QR).The experimental result shows the effectiveness of the proposed technique as well as improvements over the existing supervised feature selection algorithms.

Hybrid Tolerance Rough Set: PSO Based Supervised Feature Selection for Digital Mammogram Images

Breast cancer is the most common malignant tumor found among young and middle aged women. Feature Selection is a process of selecting most enlightening features from the data set which preserves the original significance of the features following reduction. The traditional rough set method cannot be directly applied to deafening data. This is usually addressed by employing a discretization method, which can result in information loss. This paper proposes an approach based on the tolerance rough set model, which has the flair to deal with real-valued data whilst simultaneously retaining dataset semantics. In this paper, a novel supervised feature selection in mammogram images, using Tolerance Rough Set-PSO based Quick Reduct STRSPSO-QR and Tolerance Rough Set-PSO based Relative Reduct STRSPSO-RR, is proposed. The results obtained using the proposed methods show an increase in the diagnostic accuracy.

PSO-based feature selection and neighborhood rough set-based classification for BCI multiclass motor imagery task

In recent years, most of the researchers are developing brain–computer interface (BCI) applications for the physically disabled to be able to interconnect with peripheral devices based on brain activities. Electroencephalogram (EEG) is a very powerful tool for investigating patient’s health and different physiological activities of the brain. A significant challenge in this BCI application is the accurate and reliable recognition of motor imagery (MI) task. A brain–computer interface based on MI interprets the patient’s brain activities into a control signal through classifying EEG patterns of various motor imagination tasks. The appropriate features are essential to achieving higher classification accuracy of EEG motor imagery task. For EEG signal feature extraction, wavelet transform is suitable for analysis of nonlinear time series signals. Nevertheless, the dimension of the extracted feature is huge and it may reduce the performance of classification method. Dimensionality reduction and classification play an important role in BCI motor imagery research. In this study, hybridization of particle swarm optimization (PSO)-based rough set feature selection technique is proposed for achieving a minimal set of relevant features from extracted features. The selected features are applied to the proposed novel neighborhood rough set classifier (NRSC) method for classification of multiclass motor imagery. The experimental results are delivered for nine subjects of the BCI Competition 2008 Dataset IIa to show the greater performance of the proposed algorithm. The outcome of proposed algorithms produces a higher mean kappa of 0.743 compared to 0.70 from sequential updating semi-supervised spectral regression kernel discriminant analysis. Experimental results show that the strength of the proposed PSO-rough set and NRSC algorithms outperforms the champion of the BCI Competition IV Dataset IIa and other existing research using this dataset.In recent years, most of the researchers are developing brain–computer interface (BCI) applications for the physically disabled to be able to interconnect with peripheral devices based on brain activities. Electroencephalogram (EEG) is a very powerful tool for investigating patient’s health and different physiological activities of the brain. A significant challenge in this BCI application is the accurate and reliable recognition of motor imagery (MI) task. A brain–computer interface based on MI interprets the patient’s brain activities into a control signal through classifying EEG patterns of various motor imagination tasks. The appropriate features are essential to achieving higher classification accuracy of EEG motor imagery task. For EEG signal feature extraction, wavelet transform is suitable for analysis of nonlinear time series signals. Nevertheless, the dimension of the extracted feature is huge and it may reduce the performance of classification method. Dimensionality reduction and classification play an important role in BCI motor imagery research. In this study, hybridization of particle swarm optimization (PSO)-based rough set feature selection technique is proposed for achieving a minimal set of relevant features from extracted features. The selected features are applied to the proposed novel neighborhood rough set classifier (NRSC) method for classification of multiclass motor imagery. The experimental results are delivered for nine subjects of the BCI Competition 2008 Dataset IIa to show the greater performance of the proposed algorithm. The outcome of proposed algorithms produces a higher mean kappa of 0.743 compared to 0.70 from sequential updating semi-supervised spectral regression kernel discriminant analysis. Experimental results show that the strength of the proposed PSO-rough set and NRSC algorithms outperforms the champion of the BCI Competition IV Dataset IIa and other existing research using this dataset.

Modified Soft Rough set for Multiclass Classification

Rough set theory has been applied to several domains because of its ability to handle imperfect knowledge. Most recent extension of rough set is soft rough set, where parameterized subsets of a universal set are basic building blocks for lower and upper approximations of a subset. In this paper, a new model of soft rough set, which is called modified soft rough set (MSR) where information granules are finer than soft rough sets, is applied for classification of medical data. In this paper, rough-set-based quick reduct approach is applied for selecting relevant features and MSR is applied for multiclass classification problem and the proposed work is compared with bijective soft set (BSS)-based classification, naive Bayes, and decision table classifier algorithms based on evaluation metrics.

Improved Bijective-Soft-Set-Based Classification for Gene Expression Data

One of the important problems in using gene expression profiles to forecast cancer is how to effectively select a few useful genes to build exact models from large amount of genes. Classification is also a major issue in data mining. The classification difficulties in medical area often classify medical dataset based on the outcomes of medical analysis or report of medical action by the medical practitioner. In this study, a prediction model is proposed for the classification of cancer based on gene expression profiles. Feature selection also plays a vital role in cancer classification. Feature selection techniques can be used to extract the marker genes to improve classification accuracy efficiently by removing the unwanted noisy and redundant genes. The proposed study discusses the bijective-soft-set-based classification method for gene expression data of three different cancers, which are breast cancer, lung cancer, and leukemia cancer. The proposed algorithm is also compared with fuzzy-soft-set-based classification algorithms, fuzzy KNN, and k-nearest neighbor approach. Comparative analysis of the proposed approach shows good accuracy over other methods.

Rough Set Based Feature Selection for Egyptian Neonatal Jaundice

This paper analyses rough set based feature selection methods for early intervention and prevention of neurological dysfunction and kernicterus that are the major causes of neonatal jaundice. Newborn babies develop some degree of jaundice which requires high medical attention. Improper prediction of diseases may lead to choose unsuitable type of treatment. Traditional rough set based feature selection methods and tolerance rough set based feature selection methods for supervised and unsupervised approach is applied for Egyptian neonatal jaundice dataset. Features responsible for prediction of Egyptian neonatal jaundice is analyzed using supervised quick reduct, supervised entropy based reduct and Unsupervised Tolerance Rough Set based Quick Reduct (U-TRS-QR). Results obtained demonstrate features selected by U-TRS-QR are highly accurate and will be helpful for physicians for early diagnosis.

Mining and Analysis of Clickstream Patterns

The explosive growth of the web has drastically changed the way in which information is managed and accessed. The large-scale of web data sources and the wide availability of services over the internet have increased the need for effective web data mining techniques and mechanisms . A sophisticated method to organize the layout of the information and assist user navigation is therefore particularly important. In this work, we focus on web usage mining, applying data mining techniques to web server logs. Web usage mining is the non-trivial process of distinguishing implicit, previously unknown but potentially useful clickstream patterns that may exist in any collection of web access logs. The required abstraction can be generated by clustering the web access logs based on some sort of similarity measure. Clustering is done such that the web access logs within the same group or cluster are more similar than data points from different clusters. In this chapter, we propose a partitional algorithm namely Multi Pass Combined Standard Deviation(CSD) Means algorithm which automatically generates the optimum number of clusters from the web clickstream patterns. The quality of clusters obtained using these algorithms are compared using K-Means algorithm, Rough K-Means algorithm and model based algorithms ANTCLUST and ACCANTCLUST. The experimental analysis of mined clickstream patterns shows the effectiveness of the proposed algorithm.