Quickshift++: Provably Good Initializations for Sample-Based Mean Shift

Heinrich Jiang, Jennifer Jang, Samory Kpotufe
; Proceedings of the 35th International Conference on Machine Learning, PMLR 80:2294-2303, 2018.

Abstract

We provide initial seedings to the Quick Shift clustering algorithm, which approximate the locally high-density regions of the data. Such seedings act as more stable and expressive cluster-cores than the singleton modes found by Quick Shift. We establish statistical consistency guarantees for this modification. We then show strong clustering performance on real datasets as well as promising applications to image segmentation.

Cite this Paper


BibTeX
@InProceedings{pmlr-v80-jiang18b, title = {Quickshift++: Provably Good Initializations for Sample-Based Mean Shift}, author = {Jiang, Heinrich and Jang, Jennifer and Kpotufe, Samory}, booktitle = {Proceedings of the 35th International Conference on Machine Learning}, pages = {2294--2303}, year = {2018}, editor = {Jennifer Dy and Andreas Krause}, volume = {80}, series = {Proceedings of Machine Learning Research}, address = {Stockholmsmässan, Stockholm Sweden}, month = {10--15 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v80/jiang18b/jiang18b.pdf}, url = {http://proceedings.mlr.press/v80/jiang18b.html}, abstract = {We provide initial seedings to the Quick Shift clustering algorithm, which approximate the locally high-density regions of the data. Such seedings act as more stable and expressive cluster-cores than the singleton modes found by Quick Shift. We establish statistical consistency guarantees for this modification. We then show strong clustering performance on real datasets as well as promising applications to image segmentation.} }
Endnote
%0 Conference Paper %T Quickshift++: Provably Good Initializations for Sample-Based Mean Shift %A Heinrich Jiang %A Jennifer Jang %A Samory Kpotufe %B Proceedings of the 35th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2018 %E Jennifer Dy %E Andreas Krause %F pmlr-v80-jiang18b %I PMLR %J Proceedings of Machine Learning Research %P 2294--2303 %U http://proceedings.mlr.press %V 80 %W PMLR %X We provide initial seedings to the Quick Shift clustering algorithm, which approximate the locally high-density regions of the data. Such seedings act as more stable and expressive cluster-cores than the singleton modes found by Quick Shift. We establish statistical consistency guarantees for this modification. We then show strong clustering performance on real datasets as well as promising applications to image segmentation.
APA
Jiang, H., Jang, J. & Kpotufe, S.. (2018). Quickshift++: Provably Good Initializations for Sample-Based Mean Shift. Proceedings of the 35th International Conference on Machine Learning, in PMLR 80:2294-2303

Related Material