SAGAR GUPTA
ERP Implementation Leader
ERP consultant
ERP consultant
My Reviewed Articles
Research Paper : Automation in Banking: Simplifying Operations and Enhancing Customer Experience
The banking industry is undergoing a significant transformation with the integration of automation technologies such as Artificial Intelligence (AI), Robotic Process Automation (RPA), and advanced data analytics. Automation streamlines banking operations by reducing manual intervention, increasing efficiency, and minimizing errors. AI-powered chatbots enhance customer service with instant support, while automated fraud detection systems strengthen security and compliance. Additionally, automation improves regulatory adherence by facilitating real-time monitoring and reporting, ensuring transparency and risk mitigation. The implementation of automation also leads to cost savings, operational scalability, and seamless digital banking experiences. As the industry moves towards fully automated banking ecosystems and blockchain integration, automation is set to redefine the financial landscape, making banking more accessible, secure, and customer-centric.
Research Paper : SECURING DATA TRANSFER USING PARALLEL ENCRYPTION BASED ON DIFFERENT METADATA
In the digital age, the security of multimedia data on disk drives is increasingly vital, as data volumes grow rapidly and security threats evolve. Current cryptographic methods, such as RSA-2048, though widely used, present significant limitations. RSA-2048 is computationally intensive, resulting in slow encryption and decryption speeds, which can be impractical for large volumes of multimedia data. Additionally, the RSA-2048 algorithm has vulnerabilities that may allow tampering, posing a threat to data integrity and confidentiality. These shortcomings, combined with the lack of flexibility and constant manual input required from users, make it less suitable for modern applications demanding high efficiency and robust security.
Research Paper : PHISHING WEBSITE DETECTION USING RECURRENT NEURAL NETWORKS WITH AUTOENCODERS
Phishing websites deceive users by imitating legitimate platforms to steal sensitive information, often outpacing traditional detection methods like blacklisting. These conventional approaches struggle to keep up with the rapid creation of new phishing sites, necessitating more adaptive solutions. A deep learning approach combining Recurrent Neural Networks (RNN) and autoencoders addresses this challenge effectively. RNNs, designed to process sequential data, analyze URL patterns and identify subtle temporal relationships that traditional methods miss. This capability allows RNNs to detect phishing URLs with greater accuracy, even when attackers use sophisticated evasion techniques. Autoencoders complement RNNs by performing dimensionality reduction and extracting key features from the data. This process not only enhances computational efficiency but also ensures the model focuses on the most critical aspects of phishing URLs, eliminating irrelevant noise. Together, RNNs and autoencoders form a robust detection system capable of adapting to the continuously evolving phishing landscape. This integrated model surpasses the limitations of Multilayer Perceptron (MLP) algorithms, which lack sequential data processing capabilities, and outperforms traditional methods in responsiveness and accuracy. By enabling proactive, real-time phishing detection, this approach significantly strengthens cybersecurity defenses, protecting users from the growing threat of phishing attacks.
Research Paper : Federated Learning in Ransomware Detection: A Systematic Literature Review
The exploitative and destructive challenges posed my ransomwares has continues to persist within the cyberspace industry. The increasing frequency and complexity of ransomware attacks threaten data security, resulting in substantial financial losses and operational disruptions across sectors. Recently, Federated Learning (FL) technology has been identified as a prospect for improvement in ransomware detection and mitigation. This trend is because it provides a decentralized method using machine learning (ML)/deep learning (DL) techniques to enable the collaborative training of multiple devices without providing access to their private information. This Systematic Literature Review (SLR) synthesizes the current applications of FL in ransomware detection, providing a critical evaluation of the successes and limitations of these approaches. Additionally, the review explores the evolving ransomware threat landscape and offers suggestions for future research directions to strengthen ransomware defenses. Our review began by identifying 185 relevant publications from 2019 to 2024. After thoroughly examining their abstracts, methodologies, and full texts, 53 key papers were selected for in-depth analysis. These articles were sourced from reputable databases, including Scopus, Web of Science, Springer Nature, and IEEE, among others, with the findings reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Our study addresses four critical research questions:(RQ1, RQ2, RQ3, and RQ4). Through these questions, this SLR presents a complete overview of the recent happenings in ransomware detection using FL, demonstrating valuable insights and emerging trends that can guide researchers and practitioners in crafting more effective strategies to combat ransomware attacks.
