You are on page 1of 1

Optimization Of User-Item Matrix Using SGD (Stochastic Gradient

Descent) For Movie Recommender Systems

ABSTRACT: Collaborative filtering is one of the most successful approaches to build


recommender systems. The key idea behind collaborative filtering is that similar users share
the same interest and that similar items are liked by a user. It uses the known preferences of a
group of users to make recommendations or predictions of the unknown preferences for other
users. It uses user-item matrix as an input to generate predictions of the target user. Data
sparsity is one of the major problems faced by collaborative filtering due to which the
recommendation accuracy decreases. Matrix factorization is the most commonly used
technique to overcome sparsity. Many matrix factorization techniques have been introduced
in the literature, so far, out of which Stochastic Gradient Descent algorithm is expected to
give better prediction. Stochastic gradient descent is an iterative method for optimizing an
objective function with suitable smoothness properties (e.g. differentiable or sub
differentiable). It can be regarded as a stochastic approximation, since it replaces the actual
gradient by an estimate thereof. In this paper , we use SGD algorithm on movie lens dataset.
Keywords: Collaborative filtering, Sparsity, Matrix Factorization, Stochastic Gradient
Descent.

You might also like