Resilient Backpropagation (Rprop) for Batch-learning in TensorFlow

TY - CONF

T1 - Resilient Backpropagation (Rprop) for Batch-learning in TensorFlow

AU - Florescu, Ciprian

AU - Igel, Christian

N1 - Conference code: 6

PY - 2018

Y1 - 2018

N2 - The resilient backpropagation (Rprop) algorithms are fast and accurate batch learning methods for neural networks. We describe their implementation in the popular machine learning framework TensorFlow. We present the first empirical evaluation of Rprop for training recurrent neural networks with gated recurrent units. In our experiments, Rprop with default hyperparameters outperformed vanilla steepest descent as well as the optimization algorithms RMSprop and Adam even if their hyperparameters were tuned.

AB - The resilient backpropagation (Rprop) algorithms are fast and accurate batch learning methods for neural networks. We describe their implementation in the popular machine learning framework TensorFlow. We present the first empirical evaluation of Rprop for training recurrent neural networks with gated recurrent units. In our experiments, Rprop with default hyperparameters outperformed vanilla steepest descent as well as the optimization algorithms RMSprop and Adam even if their hyperparameters were tuned.

UR - https://openreview.net/forum?id=r1R0o7yDz

M3 - Paper

T2 - International Conference on Learning Representations

Y2 - 30 April 2018 through 3 May 2018

ER -