Scalable Certified Segmentation via Randomized Smoothing

Marc Fischer, Maximilian Baader, Martin Vechev
Proceedings of the 38th International Conference on Machine Learning, PMLR 139:3340-3351, 2021.

Abstract

We present a new certification method for image and point cloud segmentation based on randomized smoothing. The method leverages a novel scalable algorithm for prediction and certification that correctly accounts for multiple testing, necessary for ensuring statistical guarantees. The key to our approach is reliance on established multiple-testing correction mechanisms as well as the ability to abstain from classifying single pixels or points while still robustly segmenting the overall input. Our experimental evaluation on synthetic data and challenging datasets, such as Pascal Context, Cityscapes, and ShapeNet, shows that our algorithm can achieve, for the first time, competitive accuracy and certification guarantees on real-world segmentation tasks. We provide an implementation at https://github.com/eth-sri/segmentation-smoothing.

Cite this Paper

BibTeX
@InProceedings{pmlr-v139-fischer21a, title = {Scalable Certified Segmentation via Randomized Smoothing}, author = {Fischer, Marc and Baader, Maximilian and Vechev, Martin}, booktitle = {Proceedings of the 38th International Conference on Machine Learning}, pages = {3340--3351}, year = {2021}, editor = {Meila, Marina and Zhang, Tong}, volume = {139}, series = {Proceedings of Machine Learning Research}, month = {18--24 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v139/fischer21a/fischer21a.pdf}, url = {https://proceedings.mlr.press/v139/fischer21a.html}, abstract = {We present a new certification method for image and point cloud segmentation based on randomized smoothing. The method leverages a novel scalable algorithm for prediction and certification that correctly accounts for multiple testing, necessary for ensuring statistical guarantees. The key to our approach is reliance on established multiple-testing correction mechanisms as well as the ability to abstain from classifying single pixels or points while still robustly segmenting the overall input. Our experimental evaluation on synthetic data and challenging datasets, such as Pascal Context, Cityscapes, and ShapeNet, shows that our algorithm can achieve, for the first time, competitive accuracy and certification guarantees on real-world segmentation tasks. We provide an implementation at https://github.com/eth-sri/segmentation-smoothing.} }
Endnote
%0 Conference Paper %T Scalable Certified Segmentation via Randomized Smoothing %A Marc Fischer %A Maximilian Baader %A Martin Vechev %B Proceedings of the 38th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2021 %E Marina Meila %E Tong Zhang %F pmlr-v139-fischer21a %I PMLR %P 3340--3351 %U https://proceedings.mlr.press/v139/fischer21a.html %V 139 %X We present a new certification method for image and point cloud segmentation based on randomized smoothing. The method leverages a novel scalable algorithm for prediction and certification that correctly accounts for multiple testing, necessary for ensuring statistical guarantees. The key to our approach is reliance on established multiple-testing correction mechanisms as well as the ability to abstain from classifying single pixels or points while still robustly segmenting the overall input. Our experimental evaluation on synthetic data and challenging datasets, such as Pascal Context, Cityscapes, and ShapeNet, shows that our algorithm can achieve, for the first time, competitive accuracy and certification guarantees on real-world segmentation tasks. We provide an implementation at https://github.com/eth-sri/segmentation-smoothing.
APA
Fischer, M., Baader, M. & Vechev, M.. (2021). Scalable Certified Segmentation via Randomized Smoothing. Proceedings of the 38th International Conference on Machine Learning, in Proceedings of Machine Learning Research 139:3340-3351 Available from https://proceedings.mlr.press/v139/fischer21a.html.

Related Material