Towards Improving the Consistency, Efficiency, and Flexibility of Differentiable Neural Architecture Search | IEEE Conference Publication | IEEE Xplore