1: \begin{abstract}

2: %In the big data era, batch based learning is more restrictive and complex for real world problems. Online learning learn models incrementally form data in a sequential manner. Most of the online learning approaches are based on the first-order online learning algorithms, which are simple and efficient but fall short in the slow convergence and sub-optimal solution. To solve this problem. we proposed a novel framework of modified newton step with element wise multiplication. The proposed algorithm achieve a minimum training time compared to other second order online learning algorithms. In this algorithm, we kept the dimension of second order as well as first order matrices and reduce the matrices size using element wise multiplication. We conduct both theoretical analysis and empirical results, including a extensive set of experiments on variety of large-scale multiclass datasets, in which the credible empirical results validate the efficiency and scalability of the proposed algorithm towards large-scale online learning algorithm.%

3: The second order method as Newton step is a suitable technique in Online Learning to guarantee regret bound. The large data is a challenge in Newton method to store second order matrices as hessian. In this paper, we have proposed a modified online Newton step that store first and second order matrices of dimension $m$ (classes) by $d$ (features). we have used element wise arithmetic operation to retain matrices size same. The modified second order matrix size results in faster computations. Also, the mistake rate is at par with respect to popular methods in literature. The experimental outcome indicates that proposed method could be helpful to handle large multi class datasets in common desktop machines using second order method as Newton step.

4: \end{abstract}