Recent Papers

The rapid growth of smart campuses calls for innovative applications that enhance student life while ensuring privacy and security. Food recommendation systems have gained popularity as personalized services to help users select suitable meals based on preferences and dietary restrictions. However, centralized data aggregation in recommendation engines raises privacy concerns, especially in campus environments where sensitive user information is involved. This paper proposes a novel Smart Campus Food Recommendation System utilizing Federated Learning (FL) combined with Privacy-Preserving Artificial Intelligence (AI) techniques. The system trains personalized recommendation models locally on user devices and aggregates model updates securely without exposing raw data, thus preserving user privacy. Differential Privacy (DP) and Secure Aggregation protocols are integrated to further protect against inference attacks. We evaluate the system on a simulated campus dataset, comparing it with traditional centralized approaches in terms of accuracy, privacy leakage, and communication efficiency. Results demonstrate that the proposed system achieves comparable recommendation quality while significantly enhancing privacy protection. The study presents a scalable, secure framework for deploying food recommendation systems in smart campus environments and lays the groundwork for future research in privacy-conscious AI-driven applications. In addition to ensuring user privacy, the system fosters user trust and complies with data protection regulations such as GDPR. Its modular architecture allows easy integration with existing campus infrastructures and scalability across institutions. The approach not only empowers individual users with control over their data but also demonstrates the practical feasibility of deploying federated learning in real-world campus applications, bridging the gap between AI innovation and ethical responsibility. Future extensions may include real-time personalization, multi-modal data inputs, and adaptive learning rates for improved responsiveness and system efficiency.

Copyright © 2025 Aakarsh Tiwari. This is an open access article distributed under the Creative Commons Attribution License.