Publication

Abstract : Due to the immense growth of information on web,it has already became a primary source of information for each and everyone in this world. As the number of users increase, the organizers update the data, and this will be readily available on the web. Many years ago, a lot of algorithms were developed in order to make retrieval easier and convenient. That is what we call generating big data, which is developed by various applications on different elds such as medical, E-Learning and networking. Therefore it is necessary that we need a better algorithms for faster performance and are cost effective computationally; hence produces more faster output. SVD on sparse representing is such solution in big data analysis. Main aim is iterative SVD update on data representation in dictionary learning. It is well known for data representation in widely applied areas of image processing,compressed sensing,etc. It has two phases i.e. sparse coding and dictionary update. Dictionary learning algorithms can be used along with pursuit algorithms like Singular Value Decomposition (SVD) in dictionary update phase and Orthogonal Matching Pursuit algorithms in sparse representation. This paper presents well implemented an iterative error update phase using Batch OMP algorithm and SVD in dictionary learning, nally we get the best updated dictionary. Our online SVD update method shows that it takes less time than the normal SVD update in dictionary learning.