Edusphere: AI and Web Integrated Platform for Cognitive Learning

Conventional Learning Management Systems (LMS) have continuously faced challenges in addressing inefficiencies in administration and managing resources effectively. This calls for the integration of traditional systems with AI. Key features include dynamic attendance systems with k- Nearest Neighbors (KNN), Convolutional Neural Networks (CNN), and Support Vector Machines (SVM), and models; personalized learning assistance provided by web scraping and automated detection of fake certificate images using CNN, CNN+LSTM, and Random Forest models. Apart from that, features such as LLM-powered chatbots, messaging systems, dynamic calendars, and interactive dashboards enhance user engagement. The results of the face recognition model indicate an accuracy of 99.37% with CNN, and 100% with both SVM and KNN models. For fake certificate image detection, the CNN model achieved 88.81% accuracy, CNN+LSTM reached $\mathbf{8 9. 1 2 \%}$, and Random Forest achieved 77.25%. This study highlights the potential of deep learning models by showcasing their performance and role in optimizing the efficiency of educational systems.