Input Switched Affine Networks: An RNN Architecture Designed for Interpretability
Jakob N. Foerster, Justin Gilmer, Jascha Sohl-Dickstein, Jan Chorowski, David Sussillo
Abstract
There exist many problem domains where the interpretability of neural network models is essential for deployment. Here we introduce a recurrent architecture composed of input-switched affine transformations – in other words an RNN without any explicit nonlinearities, but with input-dependent recurrent weights. This simple form allows the RNN to be analyzed via straightforward linear methods: we can exactly characterize the linear contribution of each input to the model predictions; we can use a change-of-basis to disentangle input, output, and computational hidden unit subspaces; we can fully reverse-engineer the architecture’s solution to a simple task. Despite this ease of interpretation, the input switched affine network achieves reasonable performance on a text modeling tasks, and allows greater computational efficiency than networks with standard nonlinearities.
BibTeX
@InProceedings{pmlr-v70-foerster17a,
title = {Input Switched Affine Networks: An {RNN} Architecture Designed for Interpretability},
author = {Jakob N. Foerster and Justin Gilmer and Jascha Sohl-Dickstein and Jan Chorowski and David Sussillo},
booktitle = {Proceedings of the 34th International Conference on Machine Learning},
pages = {1136--1145},
year = {2017},
editor = {Precup, Doina and Teh, Yee Whye},
volume = {70},
series = {Proceedings of Machine Learning Research},
month = {06--11 Aug},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v70/foerster17a/foerster17a.pdf},
url = {https://proceedings.mlr.press/v70/foerster17a.html},
abstract = {There exist many problem domains where the interpretability of neural network models is essential for deployment. Here we introduce a recurrent architecture composed of input-switched affine transformations – in other words an RNN without any explicit nonlinearities, but with input-dependent recurrent weights. This simple form allows the RNN to be analyzed via straightforward linear methods: we can exactly characterize the linear contribution of each input to the model predictions; we can use a change-of-basis to disentangle input, output, and computational hidden unit subspaces; we can fully reverse-engineer the architecture’s solution to a simple task. Despite this ease of interpretation, the input switched affine network achieves reasonable performance on a text modeling tasks, and allows greater computational efficiency than networks with standard nonlinearities.}
}