State-Regularized Recurrent Neural Networks

Cheng Wang, Mathias Niepert
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:6596-6606, 2019.

Abstract

Recurrent neural networks are a widely used class of neural architectures with two shortcomings. First, it is difficult to understand what exactly they learn. Second, they tend to work poorly on sequences requiring long-term memorization, despite having this capacity in principle. We aim to address both shortcomings with a class of recurrent networks that use a stochastic state transition mechanism between cell applications. This mechanism, which we term state-regularization, makes RNNs transition between a finite set of learnable states. We evaluate state-regularized RNNs on (1) regular languages for the purpose of automata extraction; (2) nonregular languages such as balanced parentheses, palindromes, and the copy task where external memory is required; and (3) real-word sequence learning tasks for sentiment analysis, visual object recognition, and language modeling. We show that state-regularization simplifies the extraction of finite state automata from the RNN’s state transition dynamics; forces RNNs to operate more like automata with external memory and less like finite state machines; and makes RNNs more interpretable.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-wang19j, title = {State-Regularized Recurrent Neural Networks}, author = {Wang, Cheng and Niepert, Mathias}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {6596--6606}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/wang19j/wang19j.pdf}, url = {https://proceedings.mlr.press/v97/wang19j.html}, abstract = {Recurrent neural networks are a widely used class of neural architectures with two shortcomings. First, it is difficult to understand what exactly they learn. Second, they tend to work poorly on sequences requiring long-term memorization, despite having this capacity in principle. We aim to address both shortcomings with a class of recurrent networks that use a stochastic state transition mechanism between cell applications. This mechanism, which we term state-regularization, makes RNNs transition between a finite set of learnable states. We evaluate state-regularized RNNs on (1) regular languages for the purpose of automata extraction; (2) nonregular languages such as balanced parentheses, palindromes, and the copy task where external memory is required; and (3) real-word sequence learning tasks for sentiment analysis, visual object recognition, and language modeling. We show that state-regularization simplifies the extraction of finite state automata from the RNN’s state transition dynamics; forces RNNs to operate more like automata with external memory and less like finite state machines; and makes RNNs more interpretable.} }
Endnote
%0 Conference Paper %T State-Regularized Recurrent Neural Networks %A Cheng Wang %A Mathias Niepert %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-wang19j %I PMLR %P 6596--6606 %U https://proceedings.mlr.press/v97/wang19j.html %V 97 %X Recurrent neural networks are a widely used class of neural architectures with two shortcomings. First, it is difficult to understand what exactly they learn. Second, they tend to work poorly on sequences requiring long-term memorization, despite having this capacity in principle. We aim to address both shortcomings with a class of recurrent networks that use a stochastic state transition mechanism between cell applications. This mechanism, which we term state-regularization, makes RNNs transition between a finite set of learnable states. We evaluate state-regularized RNNs on (1) regular languages for the purpose of automata extraction; (2) nonregular languages such as balanced parentheses, palindromes, and the copy task where external memory is required; and (3) real-word sequence learning tasks for sentiment analysis, visual object recognition, and language modeling. We show that state-regularization simplifies the extraction of finite state automata from the RNN’s state transition dynamics; forces RNNs to operate more like automata with external memory and less like finite state machines; and makes RNNs more interpretable.
APA
Wang, C. & Niepert, M.. (2019). State-Regularized Recurrent Neural Networks. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:6596-6606 Available from https://proceedings.mlr.press/v97/wang19j.html.

Related Material