Improving training time of deep neural networkwith asynchronous averaged stochastic gradient descent | IEEE Conference Publication | IEEE Xplore