AntisymmetricRNN: A Dynamical System View on Recurrent Neural Networks

ICLR 2019  ·  Bo Chang, Minmin Chen, Eldad Haber, Ed H. Chi ·

Recurrent neural networks have gained widespread use in modeling sequential data. Learning long-term dependencies using these models remains difficult though, due to exploding or vanishing gradients. In this paper, we draw connections between recurrent networks and ordinary differential equations. A special form of recurrent networks called the AntisymmetricRNN is proposed under this theoretical framework, which is able to capture long-term dependencies thanks to the stability property of its underlying differential equation. Existing approaches to improving RNN trainability often incur significant computation overhead. In comparison, AntisymmetricRNN achieves the same goal by design. We showcase the advantage of this new architecture through extensive simulations and experiments. AntisymmetricRNN exhibits much more predictable dynamics. It outperforms regular LSTM models on tasks requiring long-term memory and matches the performance on tasks where short-term dependencies dominate despite being much simpler.

PDF Abstract ICLR 2019 PDF ICLR 2019 Abstract


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Sequential Image Classification noise padded CIFAR-10 AntisymmetricRNN w/ gating % Test Accuracy 54.7 # 6
Sequential Image Classification noise padded CIFAR-10 LSTM % Test Accuracy 11.6 # 7