TensorFlow to Cloud FPGAs: Tradeoffs for Accelerating Deep Neural Networks | IEEE Conference Publication | IEEE Xplore