– Built a Graph-based fair recommendation model to obscure sensitive attributes in user and item embeddings, replicating the approach based on Learning Fair Representations for Recommendation: A Graph-based Perspective.
– Implemented Adversarial training techniques and utilized real-world datasets (MovieLens, Last.fm) to evaluate model performance, employing Graph Convolutional Networks (GCN), Multilayer Perceptrons (MLP), and optimization via Adam for enhancing fairness without compromising recommendation accuracy.