With the rise in the employment of deep learning methods in safety-critical scenarios, interpretability is more essential than ever before. Although many different directions regarding interpretability have been explored for visual modalities, time-series data has been neglected with only a handful of methods tested due to their poor intelligibility. We approach the problem of interpretability in a novel way by proposing TSInsight where we attach an auto-encoder to the classifier with a sparsity-inducing norm on its output and fine-tune it based on the gradients from the classifier and a reconstruction penalty. TSInsight learns to preserve features that are important for prediction by the classifier and suppresses those that are irrelevant ie serves as a feature attribution method to boost interpretability. In contrast to most other attribution frameworks, TSInsight is capable of generating both instance-based and model-based explanations. We evaluated TSInsight along with 9 other commonly used attribution methods on 8 different time-series datasets to validate its efficacy. Evaluation results show that TSInsight naturally achieves output space contraction, therefore, is an effective tool for the interpretability of deep time-series models.
@article{pub11092,
author = {
Siddiqui, Muhammad Shoaib Ahmed
and
Mercier, Dominique
and
Dengel, Andreas
and
Ahmed, Sheraz
},
title = {TSInsight: A local-global attribution framework for interpretability in time-series data},
year = {2021},
month = {11},
volume = {21},
pages = {1--16},
journal = {Sensors - Open Access Journal (Sensors)},
publisher = {MDPI}
}
German Research Center for Artificial Intelligence Deutsches Forschungszentrum für Künstliche Intelligenz
