Towards Non-saturating Recurrent Units for Modelling Long-term Dependencies

Modelling long-term dependencies is a challenge for recurrent neural networks. This is primarily due to the fact that gradients vanish during training, as the sequence length increases. Gradients can be attenuated by transition operators and are attenuated or dropped by activation functions. Canonical architectures like LSTM alleviate this issue by skipping information through a memory mechanism. We propose a new recurrent architecture (Non-saturating Recurrent Unit; NRU) that relies on a memory mechanism but forgoes both saturating activation functions and saturating gates, in order to further alleviate vanishing gradients. In a series of synthetic and real world tasks, we demonstrate that the proposed model is the only model that performs among the top 2 models across all tasks with and without long-term dependencies, when compared against a range of other architectures.

[1]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[2]  Yoshua Bengio,et al.  Gated Orthogonal Recurrent Units: On Learning to Forget , 2017, Neural Computation.

[3]  Geoffrey E. Hinton,et al.  Layer Normalization , 2016, ArXiv.

[4]  Misha Denil,et al.  Noisy Activation Functions , 2016, ICML.

[5]  Sergio Gomez Colmenarejo,et al.  Hybrid computing using a neural network with dynamic external memory , 2016, Nature.

[6]  Yann LeCun,et al.  Recurrent Orthogonal Networks and Long-Memory Tasks , 2016, ICML.

[7]  Tianqi Chen,et al.  Empirical Evaluation of Rectified Activations in Convolutional Network , 2015, ArXiv.

[8]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[9]  Yoshua Bengio,et al.  Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation , 2014, EMNLP.

[10]  Li Jing,et al.  Rotational Unit of Memory , 2018, ICLR.

[11]  Barnabás Póczos,et al.  The Statistical Recurrent Unit , 2017, ICML.

[12]  Ronald J. Williams,et al.  Simple Statistical Gradient-Following Algorithms for Connectionist Reinforcement Learning , 2004, Machine Learning.

[13]  Yoshua Bengio,et al.  Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling , 2014, ArXiv.

[14]  Alex Graves,et al.  Asynchronous Methods for Deep Reinforcement Learning , 2016, ICML.

[15]  Jürgen Schmidhuber,et al.  Learning to Forget: Continual Prediction with LSTM , 2000, Neural Computation.

[16]  Sepp Hochreiter,et al.  Untersuchungen zu dynamischen neuronalen Netzen , 1991 .

[17]  Quoc V. Le,et al.  A Neural Conversational Model , 2015, ArXiv.

[18]  Ben Poole,et al.  Categorical Reparameterization with Gumbel-Softmax , 2016, ICLR.

[19]  Alex Graves,et al.  Neural Turing Machines , 2014, ArXiv.

[20]  Yoshua Bengio,et al.  Memory Augmented Neural Networks with Wormhole Connections , 2017, ArXiv.

[21]  Aaron C. Courville,et al.  Recurrent Batch Normalization , 2016, ICLR.

[22]  Christopher Joseph Pal,et al.  On orthogonality and learning recurrent networks with long term dependencies , 2017, ICML.

[23]  Yee Whye Teh,et al.  The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables , 2016, ICLR.

[24]  Quoc V. Le,et al.  Sequence to Sequence Learning with Neural Networks , 2014, NIPS.

[25]  Richard Socher,et al.  Regularizing and Optimizing LSTM Language Models , 2017, ICLR.

[26]  Les E. Atlas,et al.  Full-Capacity Unitary Recurrent Neural Networks , 2016, NIPS.

[27]  Yoshua Bengio,et al.  Dynamic Neural Turing Machine with Continuous and Discrete Addressing Schemes , 2018, Neural Computation.

[28]  Yann Ollivier,et al.  Can recurrent neural networks warp time? , 2018, ICLR.

[29]  Yoshua Bengio,et al.  Unitary Evolution Recurrent Neural Networks , 2015, ICML.

[30]  Beatrice Santorini,et al.  Building a Large Annotated Corpus of English: The Penn Treebank , 1993, CL.

[31]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.