Hyperml and deep interest network to build a recommender system for Modatta: data privacy with gan

• Autor(a) principal: Ze, Wu Zhen
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10362/140159
• topic: Machine learning; Deep learning; Hyperbolic embeddings; Data monetization; Recommender system; Generative adversarial network; Synthetic data; Business analysis; Domínio/Área Científica::Ciências Sociais::Economia e Gestão

This work project intends to propose a privacy-based system to Modatta, a start-up focused on monetising users' data, eliminating the concerns of data leakage. The system consists of the following techniques: Deep Interest Network(DIN)/ Hyperbolic Embedding(HE), Generative Adversarial Network(GAN) and Federated Learning(FL), providing a recommender system and protecting the users' privacy. Data protection has been a hotly debated topic in society for many years, especially the adverse social effects caused by the misuse of user privacy by technology giants. This report will show that GAN is one of the feasible solutions to tackle these concerns.