TensorFlow Machine Learning Projects

Matrix factorization

Matrix factorization is a popular algorithm for implementing recommendation systems and falls in the collaborative filtering algorithms category. In this algorithm, the user-item interaction is decomposed into two low-dimensional matrices. For example, let's say all the visitor-item interactions in our dataset are M x N matrix, denoted by A. Matrix factorization decomposes matrix A into two matrices of M x k and k x N dimensions respectively, such that the dot product of these two can approximate matrix A. Some of the more popular algorithms for finding the low-dimensional matrix are based on Singular Value Decomposition (SVD). In the following example, we'll use the TensorFlow and Keras libraries to implement matrix ...