Weakly-Supervised Temporal Localization via Occurrence Count Learning

Julien Schroeter, Kirill Sidorov, David Marshall
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:5649-5659, 2019.

Abstract

We propose a novel model for temporal detection and localization which allows the training of deep neural networks using only counts of event occurrences as training labels. This powerful weakly-supervised framework alleviates the burden of the imprecise and time consuming process of annotating event locations in temporal data. Unlike existing methods, in which localization is explicitly achieved by design, our model learns localization implicitly as a byproduct of learning to count instances. This unique feature is a direct consequence of the model’s theoretical properties. We validate the effectiveness of our approach in a number of experiments (drum hit and piano onset detection in audio, digit detection in images) and demonstrate performance comparable to that of fully-supervised state-of-the-art methods, despite much weaker training requirements.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-schroeter19a, title = {Weakly-Supervised Temporal Localization via Occurrence Count Learning}, author = {Schroeter, Julien and Sidorov, Kirill and Marshall, David}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {5649--5659}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/schroeter19a/schroeter19a.pdf}, url = {https://proceedings.mlr.press/v97/schroeter19a.html}, abstract = {We propose a novel model for temporal detection and localization which allows the training of deep neural networks using only counts of event occurrences as training labels. This powerful weakly-supervised framework alleviates the burden of the imprecise and time consuming process of annotating event locations in temporal data. Unlike existing methods, in which localization is explicitly achieved by design, our model learns localization implicitly as a byproduct of learning to count instances. This unique feature is a direct consequence of the model’s theoretical properties. We validate the effectiveness of our approach in a number of experiments (drum hit and piano onset detection in audio, digit detection in images) and demonstrate performance comparable to that of fully-supervised state-of-the-art methods, despite much weaker training requirements.} }
Endnote
%0 Conference Paper %T Weakly-Supervised Temporal Localization via Occurrence Count Learning %A Julien Schroeter %A Kirill Sidorov %A David Marshall %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-schroeter19a %I PMLR %P 5649--5659 %U https://proceedings.mlr.press/v97/schroeter19a.html %V 97 %X We propose a novel model for temporal detection and localization which allows the training of deep neural networks using only counts of event occurrences as training labels. This powerful weakly-supervised framework alleviates the burden of the imprecise and time consuming process of annotating event locations in temporal data. Unlike existing methods, in which localization is explicitly achieved by design, our model learns localization implicitly as a byproduct of learning to count instances. This unique feature is a direct consequence of the model’s theoretical properties. We validate the effectiveness of our approach in a number of experiments (drum hit and piano onset detection in audio, digit detection in images) and demonstrate performance comparable to that of fully-supervised state-of-the-art methods, despite much weaker training requirements.
APA
Schroeter, J., Sidorov, K. & Marshall, D.. (2019). Weakly-Supervised Temporal Localization via Occurrence Count Learning. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:5649-5659 Available from https://proceedings.mlr.press/v97/schroeter19a.html.

Related Material