arXiv:1909.03172 [cs.LG]AbstractReferencesReviewsResources Classifications Subjects Themes Keywords training neural networks, simple two-layer convolutional neural network, two-layer convolutional neural network model, spurious local optimum, perturbed mini-batch stochastic gradient algorithms Tags conference paper Journal Information Publisher Journal Year Month Volume Number Pages DOI URL Miscellaneous Typesetting Pages Language License Submit Reset