Recommender Systems helps users to find suitable products from massively available data on the internet. The most broadly applied recommendation method is collaborative filtering, which can also be subject to shilling attacks. Profile injection occurs when malicious users insert a few bogus profiles into the user-item rating database, which alters the result of the recommendation. In this paper, the shilling attack is simulated: a Random attack, segment attack, average attack, and bandwagon attack on the movie lens dataset, focusing on users with similar interests. To build trust in the system, fake profiles must be detected. Accuracy, attack size, and filler size computations were done for each model. Several machine learning algorithms are in use to classify these fake and original profiles. Here, four Machine Learning algorithms are compared and the most efficient models are KNN, random forest, and xgboost. To get more accuracy, the ensemble model used logistic regression as a meta classifier which is more accurate than individual machine learning algorithms. Our proposed model, which is stacking an ensemble model using logistic regression as a meta-classifier, will give the best accuracy in any case.

In recent years, banks have faced a loss of customers, called churn. This degrades the reputation of banks; hence, it is important to determine the difficulties that customers face. In business, churn prediction is helpful in determining metrics such as customer retention and revenue generation for various forms of Customer Relationship Management (CRM) techniques to forecast whether a customer will exit the bank. This study aims to develop a Profit-driven Churn Prediction (PCP) model using three Machine Learning (ML) techniques: an Artificial Neural Network (ANN), a Support Vector Machine (SVM), and a Random Forest Algorithm (RFA). The PCP model is developed using a correlation-based feature selection and a highly accurate ML classifier. This prediction model is trained using previous data to categorize consumers as non-churners or future churners. Customers' behavioral and demographic features are considered reliable indicators of churn prediction. The developed PCP model is tested using a bank customer churn prediction dataset obtained from the Kaggle repository. The RFA, SVM, and ANN algorithms achieved overall accuracies of 86%, 82.3%, and 97%, respectively, for the churn dataset. The classification accuracy serves as the basis for the performance of the ML classifier. Hence, the ANN classifier is more accurate than the other classifiers in this study; they have been employed in PCP for churn prediction.

Counterfeit protection is a serious concern for businesses, as it can compromise consumer safety and damage a company's reputation. To address this issue, researchers have conducted a review of available technologies and found that blockchain-based solutions offer several advantages over traditional methods. The use of REST APIs enables secure and transparent tracking of products through the supply chain, while smart contracts facilitate automatic ownership transfers. The traceability provided by blockchain technology allows stakeholders to track a product's origin and journey, improving security and transparency. Implementing a blockchain-based supply chain management system can also improve efficiency and reduce costs associated with lost or misplaced items. This research paper advocates for businesses to invest in advanced technologies like blockchain to protect their supply chains and ensure the safety and authenticity of their products. The system described in the paper involves generating unique identification numbers (UIDs) and associating them with products during the manufacturing process with QR codes and scratchable film for consumer access, while also emphasizing the need for businesses to adopt innovative technologies to maintain competitiveness and customer trust.

Massive Open Online Courses (MOOCs) have significantly impacted the basic education industry since 2012. Online platforms enable learners to connect with the instructors present worldwide and reduce learning time by approximately 50-60%. Many universities have opted for their survival in the pandemic of COVID-19. During the pandemic, novice learners were not able to enroll in the relevant courses on these platforms, and instructors also faced challenges to satisfy their learners' needs. Each online forum has its own recommender system, and these systems only recommend courses from their own platforms. As a result, these platforms fail to satisfy the learners' educational needs and thereby increase the dropout ratio. The main objective of this study is to create a single platform for learners to search for courses from multiple platforms like Coursera, Udemy, EdX, Udacity, etc., and then recommend courses according to the learning behavior of a learner. A user profile is created in three ways, i.e., by registering, uploading their CV, or through their LinkedIn accounts. The recommender system then uses this user profile as input and recommends the relevant courses for user adaption. In this paper, demographic, content-based and collaboration-based recommender systems are used for recommendations. To validate, multi-model filtering, namely random, user-based collaboration, item-based collaboration, and matrix factorization, is used to obtain the values of the performance metrics such as RMSE, precision, and recall. On the basis of the results, the best result is obtained from user-based collaboration filtering on 6,000 dimensions of the dataset. The value of RMSE in the case of user-based collaboration filtering is 0.101, the value of precision is 0.82, and the value of recall is 0.822. Thus, the learner-centric hybrid filtering-based recommender system for MOOC platforms is implemented to enhance user adaptation.

A word or text is transliterated when it is changed from one script to another while retaining its phonetic and orthographic characteristics. Natural language processing includes transliteration, which is crucial for connecting with speakers of other languages. Since Hindi is the official language of India and there is a vast amount of content in Hindi that needs to be transliterated into English for usage on a regional and foreign scale, this paper focuses on that method. Hindi must be transliterated into English in order for speakers of other languages to understand and interact with Hindi. The present study describes a hybrid method for Hindi to English transliteration that incorporates image processing and an attention-trained model, as well as its applications in many domains. The system's performance is assessed using a dataset of images of Hindi text, and the results demonstrate that the suggested method outperforms previous transliteration systems in terms of both accuracy and speed.

Convolutional Neural Networks (CNNs) have become a powerful tool for a wide range of computer vision tasks, such as image classification, object detection, and semantic segmentation. This paper provides an overview of the fundamental concepts and architectures of CNNs, highlighting recent advancements and applications. We discuss the key components of CNNs including convolutional layers, pooling layers, and activation functions. The journey of CNN from its genesis to its evolution to the one we are familiar with today is covered. Although CNN is highly capable on its own many researchers have benefitted by hybridizing CNN with quality models. Therefore, we explore various tailor-made hybrid applications of CNN that are designed to solve very specific problems. We also discuss various innovative fields of applications of CNN and discuss the wide range of fields wherein CNN is performing surprisingly well. The paper concludes with future challenges and endeavors with respect to CNN.

Digital twin technology has gained significant attention in recent years as a promising approach for improving the performance and efficiency of various systems including vehicular networks. Vehicular networks are critical for intelligent transportation systems, providing communication and coordination among vehicles, roadside infrastructure, and other entities to enable efficient traffic management, enhanced safety, and improved driving experiences. SDN is a potential networking architecture that isolates the control plane from the data plane, enabling centralized administration and control of network resources. SDN provides programmability, flexibility, and scalability, making it well-suited for managing the complex and dynamic nature of vehicular networks. Combining digital twin technology with SDN can enable intelligent management and control of vehicular networks, leading to improved performance, enhanced reliability, and efficient resource utilization. In this paper, a novel framework is proposed that leverages digital twin technology in vehicular networks using SDN. The architecture presented integrates digital twin models with SDN controllers, enabling real-time monitoring, analysis, and control of vehicular network components. The digital twin models are used to represent virtual replicas of physical vehicular components, such as vehicles, roadside units, and traffic signals, providing a holistic view of the vehicular network's behavior and performance. The digital twin models are discussed and can be used for various vehicular network management tasks, including traffic flow optimization, congestion detection and mitigation, predictive maintenance, and incident management.