c913303f392ffc643f7240b180602652-Reviews.html

Convergence properties and relation to stochastic gradient descent have been discussed. Overall, this paper is not well written and should not be accepted in the current form. All theoretical analyses are based on the potential functions defined in line 109-111. There is no citation for this potential function and it is not explained why this function is defined in this form. Using the norm of the difference between v_n and the optimal first principal component is a more standard way to analyse the convergence rate.