Online ISSN: 2515-8260

Keywords : Support Vector Machine (SVM)


R. Parthiban; S. Usharani; D. Saravanan; D. Jayakumar; Dr.U. Palani; Dr.D. StalinDavid; D. Raghuraman

European Journal of Molecular & Clinical Medicine, 2020, Volume 7, Issue 9, Pages 2511-2530

Chronic Kidney disease (CKD) is a most predominant public health concern with increasing occurrence. CKD consists of an extensive variety of path physiological processes which will be experimental along with irregular function of kidneys and progressive decrease in Glomerular Filtration Rate (GFR).In CKD prediction various data mining methods play major important role and discovering the association among effective features in this stare canister lend a hand to detect or slow progression of this CKD disease. The information is serene from the patients’ medical records. The major intention of this effort is introducing a Hybrid Filter Wrapper Embedded (HFWE) based Feature Selection (FS) to select optimal subset of features from CKD dataset. This HFWE-FS algorithm combines the procedure of filter, wrapper and embedded algorithm. Filter algorithm is executed based on the three major functions: Relief, One- R, Gain Ratio (GR) and Gini Index (GI). Wrapper algorithm is accomplished placed on the Improved Bat Algorithm (IBA) to choose analytical Attributes from the CKD dataset. Embedded algorithm is accomplished placed on the Support Vector Machine-t-statistics (SVM-t) to choose analytical attributes. The results of all feature selection algorithms are combined and named as HFWE- FS algorithm.

Product Recommendation System for Customers through Face Recognition Using Machine Learning

Ashish Sharma; Neeraj Varshney

European Journal of Molecular & Clinical Medicine, 2020, Volume 7, Issue 4, Pages 41-44

Recommendation systems, that recommend products to customers for becoming a significant solution requires a lot of investigation. Current recommender systems will recommend products in online, but in some application fields, such as at shopping malls, online recommending system will not work. So, we recommend a facial recognition integrated recommender system to compact by the recommending products for customers at a venin store, shopping malls. Assessment result shows that this framework makes suggestion well indeed. Recommender frameworks are utilized in a collection of regions and are mainly generally professed as playlist originator for video and music organizing like Netflix, YouTube and prime videos, item recommenders for organizers as Amazon, or flip cart for internet based life stages including Face book and Twitter.

Personalized Recommendation System for Promotion of Tourists Places

L Maria Michael Visuwasam; Dr. V.P. Gladis; K Kalaiselvi; S Ananya; Krithika G

European Journal of Molecular & Clinical Medicine, 2020, Volume 7, Issue 4, Pages 1793-1802

Tour is going to be the foremost uttered word in weekends and summer breaks. So as to formulate tourism scenarios, the previous researches have provided an analysis report supported processing the downloaded images which seems to be partially accurate and time latency. The goal of our project is to verify the attractiveness of places supported location provided along-side the dataset taken such foreign tourists and tourists on-site to market the tourist places(feedback) and also to segregate the people into ages and suggesting them tourist spots. We are getting to build an internet application for our project. This concept benefits the tourists in various aspects like suggesting better places that also includes less popular but exorbitant sites.


Mrs. Swetha M S; Mr. Muneshwara M S; Dr. Chethan A. S; Mr. Shivakumara T; Dr. Anil G N

European Journal of Molecular & Clinical Medicine, 2020, Volume 7, Issue 8, Pages 5459-5466

Intrusion detection and prevention systems are widely researched areas, rightly so being an integral part of network. As with all recent computing trends, Machine Learning and Deep Learning techniques have become extremely prevalent in intrusion detection and prediction systems security. The Intrusion detection system is used to detect and notify any malware activities and try to stop them. Soft computing techniques have the ability in learning data sets which is provided and it can also categories the packets or file coming through the network or any other source as normal and abnormal. Here, we will focus more on using Support Vector Machine (SVM) and Artificial Neural Network (ANN). In the proposed method, we are using SVM and ANN algorithms for the detection of malware; the data set is processed through SVM and ANN algorithms and compares their performances with respect to accuracy metrics. Since accuracy does not give a clear picture about how well classification algorithms perform, we have also measured and compared the performances of these two algorithms using AUC score. The AUC score is a value that ranges from 0 to 1 and closest to 1 will be considered as a better one. The results show that ANN can be implemented effectively for malware detection and is comparatively better than SVM.