Department of Information Technology

Permanent URI for this communityhttps://dspace.psgrkcw.com/handle/123456789/161

Browse

Start date: 2010End date: 2019

Search Results

Now showing 1 - 10 of 68
  • No Thumbnail Available
    Item
    OPTIMIZED BOUNDARY DETECTION ALGORITHM FOR POSTAL SIGNS RECOGNITION SYSTEM USING VARIANT BASED PARTICLE SWARM INTELLIGENCE
    (IEEE, 2016-10-06) Subashini P; Krishnaveni M; Manjutha M
    Sign Language is the only mode of communication for deaf and dumb people to convey their messages. Many difficulties are faced by the hearing impaired people when they come across certain areas like Banking, Hospital and Post Office. Especially, there is no proper communication aid available in post offices to support disabled people. From available literature, it is understood that computational methods have been existing in the area of sign language recognition for hearing impaired people. These recognition system acts as an interpreter to accomplish the conversion of sign language into text or voice. This paper proposes an efficient object tracking method, that improves the performance of the video recognition system, by introducing Variant based Particle Swarm Optimization (VPSO) technique in Kalman Filter (KF) through postal video signs. The experimental results prove that VPSO based Efficient Kalman Filter (EKF) provides results better than a traditional KF.
  • No Thumbnail Available
    Item
    AN OPTIMIZED CEPSTRAL FEATURE SELECTION METHOD FOR DYSFLUENCIES CLASSIFICATION USING TAMIL SPEECH DATASET
    (IEEE, 2019-10-14) Manjutha M; Subashini P; Krishnaveni M; Narmadha V
    Speech is the most important and indispensable mode of communication between humans. In communication, the continuous flow of speech gets affected due to the interruption of emotional, panic and psychological factors that cause syllable or word repetition, prolongation and interjection. Speech dysfluency is a primary challenge for speech pathologist to isolate the normal speech from the stuttered speech. The primary objective of this paper is to propose a novel approach through optimized cepstral features selection that improves the classifiers accuracy. In this paper, Particle Swarm Optimization (PSO) and Synergistic Fibroblast Optimization (SFO) were introduced to select optimal features from conventional MFCC (Mel-Frequency Cepstrum Coefficients). The optimized cepstral features from PSO and SFO of pre-processed Tamil speech data is used to discriminate among different categories of speech signals like Normal, Moderate and Sever stutter through machine learning classification methods such as Support Vector Machine (SVM) and Naive Bayes (NB). From the experimental results, the optimal selection of cepstral features using SFO algorithm has achieved high accuracy of 96.08% employed with NB which outperforms well to the feature selection of PSO and classical MFCC. The evaluation of the proposed methodology is done by using performance metrics like sensitivity, specificity, precision, f-score and accuracy.
  • No Thumbnail Available
    Item
    ANALYSIS ON REGULARITY OF SPEECH ENERGY BASED ON OPTIMAL THRESHOLDING FOR TAMIL STUTTERING DATASET
    (IEEE, 2019-10-17) M, Manjutha; P, Subashini; M, Krishnaveni
    All over the world millions of people were affected by speech disorders in which one of the significant speech disorders is stuttering. Over the past two decade immense number of research is going on in the field of fluency disorder, and still it is necessary to enhance the analysis of stuttering disorder regional-wise. The speech signal tempo will vary with each individual where the specific fluctuation in the velocity of stutter speech is typical and it is due to the intervals in the speech rate which has a significant difference in normal stuttered speech. In this paper, Regularity of Speech Energy (RSE) was analyzed as normal, moderate and severe through Tamil speaking stuttered dataset. The analysis was done based on the energy threshold obtained during the irregular release of energy which is henceforth analyzed using optimal thresholding based on Particle Swam optimization (PSO) and Synergistic Fibroblast optimization (SFO) techniques. In order to evaluate the experimental analysis on RSE, statistical measures such as mean, standard deviation, Mean Square Error (MSE) and Root Mean Square Error (RMSE) were calculated. The experimental results of analysis on RSE have proved that stuttered speaker's signal releases low energy when compared to the normal speaker where the optimal threshold energy enhances the detection of hidden speech energy.
  • No Thumbnail Available
    Item
    DETECTION AND PREVENTION OF FRAUDULENT IN USING CREDIT CARD
    (Coimbatore Institute of Information Technology./ PSGR Krishnammal College for Women, Coimbatore, 2014-01-09) J, Maria Shyla
    Credit card transactions happen at every moment all over the world. People do their financial transactions using credit card either in online or offline. In the mean time frauds associated with it are also rising. Fraud detection involves monitoring the behavior of users in order to estimate, detect or avoid unwanted behavior .In this paper, it is shown that credit card fraud that can be detected effectively using Hidden Markov Model (HMM). HMM could be trained with Hybrid Baum-Welch and Viterbi algorithms for better efficiency in detection.
  • No Thumbnail Available
    Item
    Diagnosis of Diabetes Mellitus using Bayesian Network
    (Sri Venkateswara Eductional Trust, 2015-01) J, Maria Shyla
    Diabetes mellitus, simply referred to as diabetes, is a group of metabolic diseases.There exists more than one type of diabetes, with each type having its own risks.Diabetes is ascribed to the acute conditions under which the production and consumption of insulin is disturbed in the body which consequently leads to the increase of glucose level in the blood. Bayesian networks are considered as helpful methods for the diagnosis of many diseases. They, in fact, are probable models which have been proved useful in displaying complex systems and showing the relationships between variables in a graphic way. The advantage of this model is that it can take into account the uncertainty and can get the scenarios of the system change for the evaluation of diagnosis procedures. In this study, decision tree and Bayesian models have been compared. The results indicated that the Bayesian model is much more accurate in diabetes diagnosis.
  • No Thumbnail Available
    Item
    PPI RECONSTRUCTION TO PRIORITIZE CANDIDATE GENE WITH MULTI OBJECTIVE BAT ALGORITHM
    (Coimbatore Institute of Information Technology / Nehru Arts and Science College, 2018-11-26) J, Maria Shyla; M, Renuka Devi
    Candidate gene prioritization is the process of identifying new genes as potential candidates of being associated with a disease. A random walk-based algorithm called Random Walker on the Reliable Heterogeneous Network (RWRHN) was used to prioritize potential candidate genes for inherited disease. In this paper, the prioritization of candidate gene is improved by considering the gene similarity along with the topological similarity and phenotype similarity. Then, the similarity of genes of proteins is calculated based on gene expression data. A sub space clustering is obtained by utilizing multi objective BAT algorithm. Based on topological similarity matrix and gene similarity the PPI network is reconstructed. Finally RWRHN is applied to prioritize potential candidate genes. Thus the prioritization of candidate gene is improved with the consideration of gene similarity.
  • No Thumbnail Available
    Item
    PIONEERING METHODS FOR ENHANCING PPI AND PHENOTYPE NETWORKS FOR CANDIDATE DISEASE PRIORITIZATION
    (International Journal of Engineering and Advanced Technology - Blue Eyes Intelligence Engineering & Sciences Publication, 2019-12) J, Maria Shyla; M, Renuka Devi
    The physical contacts of high-specificity between two or more protein molecules constitute Protein-Protein Interactions (PPIs). PPI networks are modeled through graphs where node denotes proteins and edges denote interaction between proteins. The PPI network plays an important role to identify the interesting disease gene candidates. But, the PPI network usually contains false interactions. Many techniques have been proposed to reconstruct PPI network to remove false interactions and improve ranking of candidate disease. Random Walk with Restart on Diffusion profile (RWRDP) and Random Walk on a Reliable Heterogeneous Network (RWRHN) was two among them. In these methods, Gene topological similarity was incorporated with original PPI network to reconstruct new PPI network. Phenotype network was constructed by calculating similarity between gene phenotypes. The reconstructed network and phenotype networks were combined to rank candidate disease genes. However, the PPI reconstruction was fully related with the quality of protein interaction data. In order to enhance the reconstruction of PPI, a Piecewise Linear Regression (PLR) based protein sequence similarity measure and Bat Algorithm based gene expression similarity were proposed with RHN. In this paper, additional measure called Interaction Level Sub cellular Localization Score (ILSLS) is proposed to further reduce the false interaction in the reconstruction of PPI network. ILSLS is the combination of Normalized Sub cellular Localization score (NSL) and Protein Multiple Location Prediction score (PMLP). The proposed work is named as Random Walker on Optimized Trustworthy Heterogeneous Sub Cellular localization aware Network (RW-OTHSN). In order to enhance the ranking of RWOTHSN, phenotype structure is considered while construction phenotype network to rank the candidate disease genes. The phenotype structure is characterized based on h*-sequence model which identify highly discriminative signatures with only a small number of genes. This proposed work is named as Random Walker on Optimized Trustworthy Heterogeneous Sub Cellular localization and Phenotype structure aware Network (RWOTHSPN). The efficiency of the proposed methods are evaluated on PPI network database in terms of Average degree, Relative Frequency for PPI reconstruction, Number of successful predictions, precision and recall for candidate disease gene ranking.
  • No Thumbnail Available
    Item
    PRIORITIZATION OF CANDIDATE GENE ASSOCIATED WITH DISEASES IMPROVED BY RANDOM WALKER ON OPTIMIZED TRUSTWORTHY HETEROGENEOUS NETWORK
    (Jour of Adv Research in Dynamical & Control Systems, 2019-04) J, Maria Shyla; M, Renuka Devi
    Candidate gene associated with diseases could be ranked by the reconstruction of PPI Network. In current biomedical research, the prioritization of candidate gene is the most essential issue. A reliable heterogeneous network was used for candidate gene prioritization for diseases. This network was constructed by fusion of reconstructed Protein-Protein Interaction (PPI) network by topological similarity, relationship between diseases and genes of proteins and phenotype similarity network. Then, the candidate genes were prioritized by Random Walker on the Reliable Heterogeneous Network (RWRHN) which is a random walk-based algorithm. PPI network reconstruction by protein characteristic further improved the prediction accuracy of disease genes. In this paper, the prioritization of candidate gene for diseases is further improved by proposed Random Walker on Optimized Trustworthy Heterogeneous Network (RW-OTHN) which additionally considering the protein sequence similarity and gene expression profile similarity while reconstructing PPI network. The protein sequence similarity is calculated by piecewise linear regression model. The gene expression profile similarity is calculated by applying sub space clustering on high dimensional gene expression profile data. The subspace clustering is processed by multi objective BAT algorithm and K means clustering. The prioritization of candidate gene is improved with the consideration of protein sequence similarity and gene expression profile similarity in PPI network reconstruction
  • No Thumbnail Available
    Item
    ANALYSIS OF VARIOUS DATA MINING TECHNIQUES TO PREDICT DIABETES MELLITUS
    (Research India Publications, 2016-01) J, Maria Shyla; M, Renuka Devi
    Data mining approach helps to diagnose patient’s diseases. Diabetes Mellitus is a chronic disease to affect various organs of the human body. Early prediction can save human life and can take control over the diseases. This paper explores the early prediction of diabetes using various data mining techniques. The dataset has taken 768 instances from PIMA Indian Dataset to determine the accuracy of the data mining techniques in prediction. The analysis proves that Modified J48 Classifier provide the highest accuracy than other techniques.
  • No Thumbnail Available
    Item
    IDENTIFICATION OF SUBGROUPS IN A DIRECTED SOCIAL NETWORK USING EDGE BETWEENNESS AND RANDOM WALKS
    (PSGR Krishnammal College for Women, 2017-12) Sathiyakumari K; Vijaya M S
    Social networks have obtained masses hobby recently, largely because of the success of online social networking Web sites and media sharing sites. In such networks, rigorous and complex interactions occur among several unique entities, leading to huge information networks with first rate commercial enterprise ability. Network detection is an unmanaged getting to know challenge that determines the community groups based on common place hobbies, career, modules, and their hierarchical agency, the usage of the records encoded in the graph topology. Locating groups from social network is a tough mission because of its topology and overlapping of various communities. In this research, edge betweenness modularity and random walks is used for detecting groups in networks with node attributes. The twitter data of the famous cricket player is used here and network of friends and followers is analyzed using two algorithms based on edge betweenness and random walks. Also the strength of extracted communities is evaluated using on modularity score and the experiment results confirmed that the cricket player’s network is dense.