Global explanations with decision rules: a co-learning approach

Géraldin Nanfack, Paul Temple, Benoît Frénay
Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence, PMLR 161:589-599, 2021.

Abstract

Black-box machine learning models can be extremely accurate. Yet, in critical applications such as in healthcare or justice, if models cannot be explained, domain experts will be reluctant to use them. A common way to explain a black-box model is to approximate it by a simpler model such as a decision tree. In this paper, we propose a co-learning framework to learn decision rules as explanations of black-box models through knowledge distillation and simultaneously constrain the black-box model by these explanations; all of this in a differentiable manner. To do so, we introduce the soft truncated Gaussian mixture analysis (STruGMA), a probabilistic model which encapsulates hyper-rectangle decision rules. With STruGMA, global explanations can be extracted by any rule learner such as decision lists, sets or trees. We provide evidences through experiments that our framework can globally explain differentiable black-box models such as neural networks. In particular, the explanation fidelity is increased, while the accuracy of the models is marginally impacted.

Cite this Paper

BibTeX
@InProceedings{pmlr-v161-nanfack21a, title = {Global explanations with decision rules: a co-learning approach}, author = {Nanfack, G\'eraldin and Temple, Paul and Fr\'enay, Beno\^it}, booktitle = {Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence}, pages = {589--599}, year = {2021}, editor = {de Campos, Cassio and Maathuis, Marloes H.}, volume = {161}, series = {Proceedings of Machine Learning Research}, month = {27--30 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v161/nanfack21a/nanfack21a.pdf}, url = {https://proceedings.mlr.press/v161/nanfack21a.html}, abstract = {Black-box machine learning models can be extremely accurate. Yet, in critical applications such as in healthcare or justice, if models cannot be explained, domain experts will be reluctant to use them. A common way to explain a black-box model is to approximate it by a simpler model such as a decision tree. In this paper, we propose a co-learning framework to learn decision rules as explanations of black-box models through knowledge distillation and simultaneously constrain the black-box model by these explanations; all of this in a differentiable manner. To do so, we introduce the soft truncated Gaussian mixture analysis (STruGMA), a probabilistic model which encapsulates hyper-rectangle decision rules. With STruGMA, global explanations can be extracted by any rule learner such as decision lists, sets or trees. We provide evidences through experiments that our framework can globally explain differentiable black-box models such as neural networks. In particular, the explanation fidelity is increased, while the accuracy of the models is marginally impacted.} }
Endnote
%0 Conference Paper %T Global explanations with decision rules: a co-learning approach %A Géraldin Nanfack %A Paul Temple %A Benoît Frénay %B Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2021 %E Cassio de Campos %E Marloes H. Maathuis %F pmlr-v161-nanfack21a %I PMLR %P 589--599 %U https://proceedings.mlr.press/v161/nanfack21a.html %V 161 %X Black-box machine learning models can be extremely accurate. Yet, in critical applications such as in healthcare or justice, if models cannot be explained, domain experts will be reluctant to use them. A common way to explain a black-box model is to approximate it by a simpler model such as a decision tree. In this paper, we propose a co-learning framework to learn decision rules as explanations of black-box models through knowledge distillation and simultaneously constrain the black-box model by these explanations; all of this in a differentiable manner. To do so, we introduce the soft truncated Gaussian mixture analysis (STruGMA), a probabilistic model which encapsulates hyper-rectangle decision rules. With STruGMA, global explanations can be extracted by any rule learner such as decision lists, sets or trees. We provide evidences through experiments that our framework can globally explain differentiable black-box models such as neural networks. In particular, the explanation fidelity is increased, while the accuracy of the models is marginally impacted.
APA
Nanfack, G., Temple, P. & Frénay, B.. (2021). Global explanations with decision rules: a co-learning approach. Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 161:589-599 Available from https://proceedings.mlr.press/v161/nanfack21a.html.

Related Material