Extending Deep Neural Categorisation Models for Recommendations by Applying Gradient Based Learning
Sashank Sridhar, Sowmya Sanagavarapu · 2021
A large number of websites hosted on the Internet use categorizations systems for the organization and storage of their big data. Big data management is performed by the design of these systems extensively ranges from using probability models to deep neural networks. Deep learning categorization models optimize the learned activation weights stored in their system for all their tasks, to reduce loss and increase the accuracy of predictions, which can be extracted and used for building recommendation systems for users on their platform for personalization and ease of use. In this paper, we have used the popular MovieLens 100k dataset for building a movie recommendation model by using the weighted class activation maps obtained from a neural classification model using cosine similarity and visualized using k-NN algorithm. The model performed with a precision and recall of 0.70 and 0.89 while giving 25 recommendations to the user for their selected movie.