Probabilistic spatial transformer networks

Pola Schwöbel, Frederik Rahbæk Warburg, Martin Jørgensen, Kristoffer Hougaard Madsen, Søren Hauberg
Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence, PMLR 180:1749-1759, 2022.

Abstract

Spatial Transformer Networks (STNs) estimate image transformations that can improve downstream tasks by ‘zooming in’ on relevant regions in an image. However, STNs are hard to train and sensitive to mis-predictions of transformations. To circumvent these limitations, we propose a probabilistic extension that estimates a stochastic transformation rather than a deterministic one. Marginalizing transformations allows us to consider each image at multiple poses, which makes the localization task easier and the training more robust. As an additional benefit, the stochastic transformations act as a localized, learned data augmentation that improves the downstream tasks. We show across standard imaging benchmarks and on a challenging real-world dataset that these two properties lead to improved classification performance, robustness and model calibration. We further demonstrate that the approach generalizes to non-visual domains by improving model performance on time-series data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v180-schwobel22a, title = {Probabilistic spatial transformer networks}, author = {Schw{\"o}bel, Pola and Warburg, Frederik Rahb{\ae}k and J{\o}rgensen, Martin and Madsen, Kristoffer Hougaard and Hauberg, S{\o}ren}, booktitle = {Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence}, pages = {1749--1759}, year = {2022}, editor = {Cussens, James and Zhang, Kun}, volume = {180}, series = {Proceedings of Machine Learning Research}, month = {01--05 Aug}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v180/schwobel22a/schwobel22a.pdf}, url = {https://proceedings.mlr.press/v180/schwobel22a.html}, abstract = {Spatial Transformer Networks (STNs) estimate image transformations that can improve downstream tasks by ‘zooming in’ on relevant regions in an image. However, STNs are hard to train and sensitive to mis-predictions of transformations. To circumvent these limitations, we propose a probabilistic extension that estimates a stochastic transformation rather than a deterministic one. Marginalizing transformations allows us to consider each image at multiple poses, which makes the localization task easier and the training more robust. As an additional benefit, the stochastic transformations act as a localized, learned data augmentation that improves the downstream tasks. We show across standard imaging benchmarks and on a challenging real-world dataset that these two properties lead to improved classification performance, robustness and model calibration. We further demonstrate that the approach generalizes to non-visual domains by improving model performance on time-series data.} }
Endnote
%0 Conference Paper %T Probabilistic spatial transformer networks %A Pola Schwöbel %A Frederik Rahbæk Warburg %A Martin Jørgensen %A Kristoffer Hougaard Madsen %A Søren Hauberg %B Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2022 %E James Cussens %E Kun Zhang %F pmlr-v180-schwobel22a %I PMLR %P 1749--1759 %U https://proceedings.mlr.press/v180/schwobel22a.html %V 180 %X Spatial Transformer Networks (STNs) estimate image transformations that can improve downstream tasks by ‘zooming in’ on relevant regions in an image. However, STNs are hard to train and sensitive to mis-predictions of transformations. To circumvent these limitations, we propose a probabilistic extension that estimates a stochastic transformation rather than a deterministic one. Marginalizing transformations allows us to consider each image at multiple poses, which makes the localization task easier and the training more robust. As an additional benefit, the stochastic transformations act as a localized, learned data augmentation that improves the downstream tasks. We show across standard imaging benchmarks and on a challenging real-world dataset that these two properties lead to improved classification performance, robustness and model calibration. We further demonstrate that the approach generalizes to non-visual domains by improving model performance on time-series data.
APA
Schwöbel, P., Warburg, F.R., Jørgensen, M., Madsen, K.H. & Hauberg, S.. (2022). Probabilistic spatial transformer networks. Proceedings of the Thirty-Eighth Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 180:1749-1759 Available from https://proceedings.mlr.press/v180/schwobel22a.html.

Related Material