Movie Recommendation System
I developed a deep learning model for a recommendation system as a task at iCog Labs, using TensorFlow and Keras. The model leverages embeddings for both users and items to capture latent factors and predict ratings. It features user and item embeddings, a dot product interaction between these embeddings, and a dense neural network layer for additional feature processing. I applied regularization to the embeddings to prevent overfitting and used a custom learning rate schedule and early stopping during training to optimize performance. I monitored the model’s performance with mean squared error and mean absolute error, performing regular validation checks to ensure its generalization capability.
📒Explore the Notebook