Active Deep Probabilistic Subsampling

Hans Van Gorp, Iris Huijben, Bastiaan S Veeling, Nicola Pezzotti, Ruud J. G. Van Sloun
Proceedings of the 38th International Conference on Machine Learning, PMLR 139:10509-10518, 2021.

Abstract

Subsampling a signal of interest can reduce costly data transfer, battery drain, radiation exposure and acquisition time in a wide range of problems. The recently proposed Deep Probabilistic Subsampling (DPS) method effectively integrates subsampling in an end-to-end deep learning model, but learns a static pattern for all datapoints. We generalize DPS to a sequential method that actively picks the next sample based on the information acquired so far; dubbed Active-DPS (A-DPS). We validate that A-DPS improves over DPS for MNIST classification at high subsampling rates. Moreover, we demonstrate strong performance in active acquisition Magnetic Resonance Image (MRI) reconstruction, outperforming DPS and other deep learning methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v139-van-gorp21a, title = {Active Deep Probabilistic Subsampling}, author = {Van Gorp, Hans and Huijben, Iris and Veeling, Bastiaan S and Pezzotti, Nicola and Van Sloun, Ruud J. G.}, booktitle = {Proceedings of the 38th International Conference on Machine Learning}, pages = {10509--10518}, 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/van-gorp21a/van-gorp21a.pdf}, url = {https://proceedings.mlr.press/v139/van-gorp21a.html}, abstract = {Subsampling a signal of interest can reduce costly data transfer, battery drain, radiation exposure and acquisition time in a wide range of problems. The recently proposed Deep Probabilistic Subsampling (DPS) method effectively integrates subsampling in an end-to-end deep learning model, but learns a static pattern for all datapoints. We generalize DPS to a sequential method that actively picks the next sample based on the information acquired so far; dubbed Active-DPS (A-DPS). We validate that A-DPS improves over DPS for MNIST classification at high subsampling rates. Moreover, we demonstrate strong performance in active acquisition Magnetic Resonance Image (MRI) reconstruction, outperforming DPS and other deep learning methods.} }
Endnote
%0 Conference Paper %T Active Deep Probabilistic Subsampling %A Hans Van Gorp %A Iris Huijben %A Bastiaan S Veeling %A Nicola Pezzotti %A Ruud J. G. Van Sloun %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-van-gorp21a %I PMLR %P 10509--10518 %U https://proceedings.mlr.press/v139/van-gorp21a.html %V 139 %X Subsampling a signal of interest can reduce costly data transfer, battery drain, radiation exposure and acquisition time in a wide range of problems. The recently proposed Deep Probabilistic Subsampling (DPS) method effectively integrates subsampling in an end-to-end deep learning model, but learns a static pattern for all datapoints. We generalize DPS to a sequential method that actively picks the next sample based on the information acquired so far; dubbed Active-DPS (A-DPS). We validate that A-DPS improves over DPS for MNIST classification at high subsampling rates. Moreover, we demonstrate strong performance in active acquisition Magnetic Resonance Image (MRI) reconstruction, outperforming DPS and other deep learning methods.
APA
Van Gorp, H., Huijben, I., Veeling, B.S., Pezzotti, N. & Van Sloun, R.J.G.. (2021). Active Deep Probabilistic Subsampling. Proceedings of the 38th International Conference on Machine Learning, in Proceedings of Machine Learning Research 139:10509-10518 Available from https://proceedings.mlr.press/v139/van-gorp21a.html.

Related Material