Evaluating the power efficiency of deep learning inference on embedded GPU systems | IEEE Conference Publication | IEEE Xplore