Hyperparameter tuning in Deep Learning TechniquesforCollaborative Recommendation

Abstract

Recommender systems are virtual devices that help people cope with information overload in a variety of fields. Customers are recommended items after a vast volume of data is analyzed to determine their preferences. Deep learning makes it possible to train models more precisely, which is challenging in a traditional environment. Later on, it necessitates a significant computational cost and has an impact on the success of big data appeals. In this article, we present a unique recommender scheme for Movielens data. It has a Multi-Layer Perceptron built in to handle large data sets and increase prediction accuracy by addressing data sparsity and scalability concerns. This dissertation focuses on the prediction model when dealing with this dataset. We discovered that our recommended model outperformed existing recommendation models when comparing Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE).