Browse Items (2150 total)
Sort by:
-
An Approach for Credit Card Churn Prediction Using Gradient Descent
A very important asset for any company in the business sectors such as banking, marketing, etc. are its customers. For them to stay in the game, they have to satisfy their customers. Customer retention plays an important role in attracting and retaining the customers. Customer retention means to keep the customer satisfied so that they do not stop using their service/product in the domain of banking; the banks provide various kinds of services to the customers especially in the electronic banking sector. For this study, we have selected the service of credit card. For a bank to give a loan or amount on credit basis, the e-bank should make sure if its customers are eligible and can repay their money. The purpose of this project is to implement a neural network model to classify the churners and non-churners. 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. -
An Approach for Detecting Frauds in E-Commerce Transactions using Machine Learning Techniques
This paper is primarily focused on E-commerce fraud detection using machine learning techniques. There are many different ways to detect E-commerce fraud using machine learning approach. In this work, comparison study is conducted between various available machine learning algorithms to detect the online frauds. During the comparative study, focus is underlined on comparison of all the algorithms to identify the fraud transactions. When compared to other algorithms, such as support vector machine, Decision Tree, K-nearest neighbour and Random Forest, it has been observed that Logistic regression gives better result among all machine learning algorithms. 2021 IEEE. -
An approach for document pre-processing and K Means algorithm implementation
The web mining is a cutting edge technology, which includes information gathering and classification of information over web. This paper puts forth the concepts of document pre-processing, which is achieved by extraction of keywords from the documents fetched from the web, processing it and generating a term-document matrix, TF-IDF and the different approaches of TF-IDF (term frequency Inverse document frequency) for each respective document. The last step is the clustering of these results through K Means algorithm, by comparing the performance of each approach used. The algorithm is realized on an X64 architecture and coded on Java and Matlab platform. The results are tabulated. 2014 IEEE. -
An approach for efficient capacity management in a cloud
Cloud computing is an emerging technology where computing resources such as software and hardware are accessed over the internet as a service to customers. In the past, due to less demand, cloud capacity management was not critical. However, with the increase in demand, capacity management has become critical. Cloud customers can frequently use web-based portals to provision and de-provision virtual machines on demand. Due to dynamic changes as per the demand, managing capacity becomes a challenging task. In this paper, we discuss the emergence of cloud computing, traditional versus cloud computing, and how capacity management can be efficiently handled in a cloud. A detail on high availability of virtual machines in a cloud using the N+1 model is discussed in this paper. With templates, many repetitive installation and configuration tasks can be avoided. We discuss the sizing of templates and the overheads of using virtual machines. We suggest ideal combinations of sizing templates to create virtual machines with optimum utilization of blades. Finally we discuss a few benefits of efficient capacity management in cloud computing. 2017 IEEE. -
An approach to improvise recognition rate from occluded and pose variant faces
Face recognition is increasingly gaining popularity in today's field mainly because one of the major applications of face recognition, surveillance cameras are being used in real world applications. At the same time, researchers are trying to increase the accuracy of recognition as recognizing face from an unconstrained faces is naturally difficult. In the case of real world application, during image capture there are high chances of faces appearing with different poses, face subjected to illumination and occlusion. In this paper we propose a model that can increase the recognition rate with faces of different pose and faces subjected to occlusion. We introduce the technique of in-painting to restore the occluded face in a frame of video. A dictionary set is created with restored occluded face and faces with varying inclination. In our proposed model, Discrete Curvelet Transform is used to extract features. Comparison with traditional method shows a better recognition rate. 2015 IEEE. -
An Approach to Introduce Mobile Application Development for Teaching and Learning by Adapting Allans Dual Coding Theory
This paper reveals around methodology that could be powerful to teach mobile applications in a class by including a decent utilization of innovative technology alongside a system. Appraisal instruments within the cloud were utilized to encourage this sort of methodology toward teaching application development. The new methodology is executed by teaching in the lab with desktops or in the classroom with students laptops. It adapted Allans dual code Theory. The adequacy of this methodology is obvious through an examination of outcomes. 2020, Springer Nature Switzerland AG. -
An Architecture for Risk-Based Authentication System in a Multi-Server Environment
Identity authentication, a vital part of any application access, is also one way for imposters to gain access to an application using various fingerprint authentication technologies. Therefore, because of the lack of security in the authentication architecture, this paper proposes an architecture for a risk-based authentication system using a machine learning model in a multi-server environment. Since the recent study mainly focuses on the multi-server environment and adaptive authentication independently, very little work has been proposed using a multi-server environment for adaptive authentication. The study aims to estimate risk for the user during the initial login process and when the user's data is extracted enough for prediction in a multi-server environment. 2023 IEEE. -
An Automated Deep Learning Model for Detecting Sarcastic Comments
The concept of Natural Language Processing is immensely vast with a wide range of fields in which ideas can be explored and innovations can be developed. An algorithm based on deep learning is used to detect sarcasm in text in this paper. It is usually only possible to detect sarcasm through speech and very rarely through text. 1.3 million comments from Reddit were analyzed, of which half were sarcastic and half were not, and then various deep learning models were applied, such as standard neural networks, CNNs, and LSTM RNNs. The best performing model was LSTM-RNNs, followed by CNNs, and standard neural networks came last. With textual data, it is much harder to understand whether the other person is being sarcastic or not, it can only be understood by listening to their tone of voice or looking at their behaviour. The purpose of this paper is to demonstrate how to detect sarcasm in textual data using deep learning models. 2021 IEEE. -
An autonomic computing architecture for business applications
Though the vision of autonomic computing (AC) is highly ambitious, an objective analysis of autonomic computing and its growth in the last decade throw more incisive and decisive insights on its birth deformities and growth pains. Predominantly software-based solutions are being preferred to make IT infrastructures and platforms, adaptive and autonomic in their offerings, outputs, and outlooks. However the autonomic journey has not been as promising as originally envisaged by industry leaders and luminaries, and there are several reasons being quoted by professionals and pundits for that gap. Precisely speaking, there is a kind of slackness in articulating its unique characteristics, and the enormous potentials in business and IT acceleration. There are not many real-world applications to popularize the autonomic concept among the development community. Though, some inroads has been made into infrastructure areas like networking, load balancing etc., very few attempts has been exercised in application areas like ERP, SCM, or CRM. In this paper, we would like to dig and dive deeper to extract and explain where the pioneering and path-breaking autonomic computing stands today, and the varied opportunities and possibilities, which insists hot pursuit of the autonomic idea. A simplistic architecture for deployment of autonomic business applications is introduced and a sample implementation in an existing CRM system is described. This should form the basis of new start and ubiquitous application of AC concepts for business applications. 2012 IEEE. -
An Early Detection of Autism Spectrum Disorder Using PDNN and ABIDE I&II Dataset
The current study's objective was to use deep learning methods to separate valetudinarians amidst autism spectrum disorders (ASDs) from controls employing just the patients brain activation patterns from a dataset of large brain images. We examined brain imaging data from ASD patients from the global, multi-site ABIDE dataset (Autism Brain Imaging Data Exchange). Social impairments and repetitive behaviors are hallmarks of the brain condition known as autism spectrum disorder (ASD). ASD affects one in every 68 kids in the USA, as of the most recent data from the Disease Control Centers. To understand the neurological patterns that arose from the categorization, we looked into functional connectivity patterns that can be used to diagnose ASD participants precisely. The outcomes raised the state of the art by correctly identifying 72.10% of ASD patients in the sample vs. control patients. The classification patterns revealed an anti-correlation between the function of the brain's anterior and posterior regions; this anti-correlation supports the empirical data currently showing achingly ASD impedes communication between the livid brain's anterior and posterior areas. We found and pinpointed brain regions damn frolic, distinguishing ASD among typically developing reign according to our deep learning model. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd 2024. -
An Early-Stage Diabetes Symptoms Detection Prototype using Ensemble Learning
Diabetes is one of the most increasing health issues that the whole world is facing. Recent research has shown that diabetes is spreading quickly in India. Having more than 77 million sufferers, India is actually regarded as the diabetes capital of the world. The lifestyle and eating patterns of people who move from rural to urban settings alter, which raises the prevalence of diabetes. Diabetes has been linked to consequences like vision loss, renal failure, nerve damage, cardiovascular disease, foot ulcers, and digestive issues. Diabetes can harm the blood arteries and neurons in a variety of organs. FPG (Flaccid Plasma Glucose) is a popular test that is done to find out whether a person is a diabetic patient or not. However, not all people consistently take medication and neither monitor their blood sugar levels on a regular basis. Early detection of this disease is also an important thing that people usually don't do. Technology these days has emerged a lot in the healthcare zone. Many prototypes have already been made for the detection of diabetes. The prototype discussed in this paper is an ensemble learning approach for the detection of diabetes in a very early stage. Ensemble learning which includes the use of multiple model prediction has been used to make the outcome stronger and more trustworthy. The overall accuracy achieved by the model is 96.54%. XGBoost also records the minimal execution time of 2.77 seconds only. 2023 IEEE. -
An Econometric Approach Towards Exploring the Impact of Workers Remittances on Inflation: Empirical Evidence from India
This paper attempts to study short and long run impact of increased workers remittances on general price level. It uses the real GDP growth, real effective exchange rate (REER), M3 (broad money), fiscal deficit to gauge the impact of foreign remittances on inflation. The study makes use of VAR/VECM framework to gauge the impact of workers remittances on inflation. Inflation is measured in terms of CPI and WPI, real income or GDP at constant prices is taken as a measure of GDP growth, REER is used for exchange rates and M3 is taken as a proxy for money supply. Monthly data of all these variables has been taken from Bloomberg and World Bank data base. The findings provide important insights into the nature of association between remittances and inflation suggesting causality between inflation, remittances, real GDP, real effective exchange rates and money supply due to increased workers remittances. The findings have policy implications for decisions to channelize workers remittances in a way to increase real GDP growth and money supply while at the same time not causing the general price levels to soar. The present study focuses on how increased (decreased) workers remittances is leading to increase (decrease) in general price levels in India. 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. -
An Effecient Approach to Detect Fraud Instagram Accounts Using Supervised ML Algorithms
Nowadays social media plays a vital role in different fields including business, economic communication and personal. Many person get profit from the different origins of availability of data from these social media, but cyber-crimes are increasing day by day. A person can generate many fake accounts and hence pretenders can easily be made. Instagram, as one of the popular types of online social media site, carries big information and messages through the posts. Most of the person use Instagram as a digital life marketing place because it is a one of the big social media site. The goal of the research paper is to recognize and stop fake IDs and pages. Because through the professional pages of Instagram, many fake cases and things are occurring present days. So the main thing is to recognize fake pages and fake accounts also. In this paper, we work on various IDs of Instagram. We want to observe an ID is real or not using Machine Learning techniques namely Logistic Regression, Naive Bayes, Support vector machine, Decision tree, Random Forest. 2022 IEEE. -
An effective Approach for Pneumonia Detection using Convolution Vision Transformer
Early detection of pneumonia in patients through effective medical imaging may enable timely remedial measures and reduce the severity of the infection. There is an increase in cases among new-borns, teenagers and also people with health issues in recent years. The COVID-19 pandemic also revealed the major impact pneumonia had on the lungs and the consequences of delayed detection. The presence of the infection in the lungs is examined through images of Chest X-ray, however, for an early diagnosis of the infection, this paper proposes an automated model as a more effective alternative. Convolutional Vision Transformer (CVT) which gives an accuracy of 97.13%, and is a robust combination of Convolution and Vision Transformer (ViT), is suggested in this paper as a potential model to detect pneumonia early in patients. 2022 IEEE. -
An Effective BiLSTM-CRF Based Approach to Predict Student Achievement: An Experimental Evaluation
Currently, massive volumes of data are accumulated in databases when people configure new requirements and services. Data mining techniques and intelligent systems are emerging for managing large amounts of data and extracting actionable insights for policy development. As digital technology has grown, it has naturally become intertwined with e-learning practices. In order to facilitate communication between instructors and a diverse student body located all over the world, distance learning programs rely on Learning Management Systems (LMSs). Colleges can better accommodate their students' individual needs by using and analyzing interaction data that reveals variances in their learning progress. Predicting pupils' success or failure is a breeze with the help of learning analytics tools. Better learning outcomes might be achieved through early prediction leading to swift focused action. Preprocessing, feature selection, and model training are the three components of the proposed method. Data cleansing, data transformation, and data reduction are the preprocessing steps used here. It used a CFS to enable feature selection. This study has used a BiLSTM-CRF hybrid approach to train the model. When compared to tried-and-true techniques like CNN and CRF, the proposed method performs effectively. 2024 IEEE. -
An Effective Deep Learning Classification of Diabetes Based Eye Disease Grades: An Retinal Analysis Approach
Diabetic Retinopathy (DR) is a common misdiagnosis of diabetes mellitus, which damages the retina and impairs eyesight. It can lead to vision impairment if it is not caught early. Tragically, DR is an unbreakable cycle, and treatment only serves to reinforce the perception. Early detection of DR and effective treatment can significantly lower the risk of visual loss. In comparison to PC-aided conclusion frameworks, the manual analysis process used by ophthalmologists to diagnose DR retina fundus images takes a lot of time, effort, and money and is prone to error. As of late, profound learning has become quite possibly the most well-known procedure that has accomplished better execution in numerous areas, particularly in clinical picture examination and classification. Thereby, this paper brings an effective deep learning-based diabetes-based retinography in which the following are the stages: a) Data collection from MESSIDOR which contains 1200 images classified into 4 levels and graded from 03 followed by b) Preprocessing using grayscale normalized data. Then followed by c) feature extraction using Discrete Wavelet Transform (DWT), d) feature selection using Particle Swarm Optimization (PSO) and finally given for e) classification using Densenet 169. Experimental states that the proposed model outperforms and effectively classified grades compared to other state-of-art models (accuracy:0.95, sensitivity:0.96, specificity;0.97). 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG. -
An Effective Time Series Analysis for Equity Market Prediction Using Deep Learning Model
A stock Exchange is a market where securities are traded. Every day, billions are traded at various stock exchanges across the world. In recent years prediction of movement of stock market is regarded as fascinating and has created a demand in financial market time series prediction. A precise forecasting of equity market is needed to provide higher returns for investors. Since there is high complexity in predicting stock market profits, developing models for it becomes difficult. The data mining and machine learning techniques has played an important role in Prediction of stock market movement. This study attempted to develop a deep learning model using Recurrent Neural Network for forecasting movement in the National Stock Exchange of India's benchmark broad based stock market index(NIFTY 50) for the Indian equity market. In this paper the NIFTY 50 index and INFYOSYS Ltd historical data from Yahoo finance companies has been selected for forecasting and analysis. 2019 IEEE. -
An Efficient and Robust Explainable Artificial Intelligence for Securing Smart Healthcare System
The advent of IoT technologies has a tremendous impact on the healthcare sector enabling efficient monitoring of patients and utilizing the data for better analytics. Since every activity related to a patients health is monitored, the focus on smart healthcare applications has significantly transferred from service provision to a security perspective. As most healthcare applications are automated security plays a vital role. The technique of machine learning has been widely used in securing smart healthcare systems. The major challenge is that these applications require high-quality labeled images, which are difficult to acquire from real-time security applications. Further, it highly time-consuming and cost-expensive process. To address these constraints, in this paper, we define an efficient and robust explainable artificial intelligence technique that takes a small quantity of labeled data to train and de-ploy the security countermeasure for targeted healthcare applications. The proposed approach enhances the security measure through the detection of drifting samples with explainability. It is observed that the proposed approach improved accuracy, high fidelity, and explanation measures. Also, this approach is proven to be considerably resistant against numerous security threats. 2023 IEEE. -
An Efficient andOptimized Convolution Neural Network forBrain Tumour Detection
Brain tumour is a life threatening disease and can affect children and adults. This study focuses on classifying MRI scan images of brain into one of 4 classes namely: glioma tumour, meningioma tumour, pituitary tumour and normal brain. Person affected with brain tumours will need treatments such as surgery, radiation therapy or chemotherapy. Pretrained Convolution Neural Networks such as VGG19, MobileNet, and AlexNet which have been widely used for image classification using transfer learning. However due to huge storage space requirements these are not effectively deployed on edge devices for creation of robotic devices. Hence a compressed version of these models have been created using Genetic Algorithm algorithm which occupies nearly 3040% of space and also a reduced inference time which is less by around 50% of original model. The accuracy provided by VGG19, AlexNet, MobileNet and Proposed CNN before compression was 92.18%, 89.45%, 93.75% and 96.85% respectively. Similarly the accuracy after compression for VGG19, AlexNet, MobileNet and Proposed CNN was 91.34%, 88.92%, 94.40% and 95.29%. 2023, Springer Nature Switzerland AG. -
An Efficient Approach for Obstacle Avoidance and Navigation in Robots
Reinforcement learning has emerged as a prominent technique for enhancing robot obstacle avoidance capabilities in recent years. This research provides a comprehensive overview of reinforcement learning methods, focusing on Bayesian, static, dynamic policy, Deep Q-Learning (DQN) and extended dynamic policy algorithms. In the context of robot obstacle avoidance, these algorithms enable an agent to interact with its physical environment, learns effective operating strategies, and optimize actions to maximize a reward signal. The environment typically consists of a physical space that the robot must navigate without encountering obstacles. The reward signal serves as an objective measure of the robot's performance towards accomplishing specific goals, such as reaching designated positions or completing tasks. Furthermore, successful obstacle avoidance strategies acquired in simulation environments can be seamlessly transferred to real-world scenarios. The promising results achieved thus far indicate the potential of reinforcement learning as a powerful tool for enhancing robot obstacle avoidance. This research concludes with insights into the future prospects of reward learning, high-lighting its ongoing importance in the development of intelligent robotics systems. The proposed algorithm DQN outperforms well among all the other algorithms with an accuracy of 81%, Through this research, we aim to provide valuable insights and directions for further advancements in the field of robot obstacle avoidance using reinforcement learning techniques. 2023 IEEE.