Recommender System @Samsung Research

Built a Collaborative Filtering-based Recommender System, which uses custom-engineered features to represent items in a classified dataset. These features are passed into a neural network made in PyTorch to generate predictions. Recommendation feedback is incorporated with gamma value similar to reward in Reinforcement Learning to vary weight given to a recommendation with time. This model was integrated with a DAPP and Blockchain Network to prioritize user data security. The model was proposed to run in three different phases to deal with cold-start issues.