Federated Learning for Secure Customer Behavior in IoT E-Commerce

The advancement of IoT ecosystems and e-commerce has made it important to study customer behavior to improve business practices and personalize offers and experiences. AI models have increased customer retention due to the use of real-time data like history of customer browsing. However, privacy concerns due to the storage of sensitive data impede the use of centralized machine learning. Federated Learning (FL) offers a different approach by allowing machine learning models to be trained on client devices for privacy-protecting data. In this paper, we analyze the personalization of e-commerce recommendations and privacy using FL in systems such as multi-view Fed-FR-MVD and IoT-Enhanced Hybrid Deep Collaborative Transformers (HDCT). Furthermore, this paper addresses the use of Differential Privacy and Homomorphic Encryption (or the two combined with blockchain) to secure data aggregation and maintain data integrity. FL has been shown to address the above issues, securing AI systems for e-commerce, by improving the accuracy and relevance of recommendations and ensuring user privacy with enhanced functionality.

کلیدواژه ها:

Federated Learning ، IoT ، E-Commerce ، Customer Behavior Analysis ، Privacy-Preserving AI ، Edge Computing