Extracting Automata from Recurrent Neural Networks Using Queries and Counterexamples
Abstract
We present a novel algorithm that uses exact learning and abstraction to extract a deterministic finite automaton describing the state dynamics of a given trained RNN. We do this using Angluin’s \lstar algorithm as a learner and the trained RNN as an oracle. Our technique efficiently extracts accurate automata from trained RNNs, even when the state vectors are large and require fine differentiation.
Cite
Text
Weiss et al. "Extracting Automata from Recurrent Neural Networks Using Queries and Counterexamples." International Conference on Machine Learning, 2018.Markdown
[Weiss et al. "Extracting Automata from Recurrent Neural Networks Using Queries and Counterexamples." International Conference on Machine Learning, 2018.](https://mlanthology.org/icml/2018/weiss2018icml-extracting/)BibTeX
@inproceedings{weiss2018icml-extracting,
title = {{Extracting Automata from Recurrent Neural Networks Using Queries and Counterexamples}},
author = {Weiss, Gail and Goldberg, Yoav and Yahav, Eran},
booktitle = {International Conference on Machine Learning},
year = {2018},
pages = {5247-5256},
volume = {80},
url = {https://mlanthology.org/icml/2018/weiss2018icml-extracting/}
}