Towards Native Execution of Deep Learning on a Leadership-Class HPC System | IEEE Conference Publication | IEEE Xplore