Stochastic smoothing can be used to design online algorithms such for problems such as variance minimization and online PCA. In the case of smoothing spectral function of matrices, the performance of the algorithm is controlled by its Hessian. I will show how to obtain a concise formula of Hessian of the largest eigenvalue of symmetric matrices and give an upper bound of the regret when we use Gaussian Orthogonal Matrix to smooth the function. Speaker(s): Yitong Sun (UM)