Quickshift++: Provably Good Initializations for Sample-Based Mean Shift
This work addresses clustering stability and expressiveness for data analysis, but it is incremental as it modifies an existing algorithm.
The paper tackled the problem of improving Quick Shift clustering by providing initial seedings that approximate high-density regions, resulting in more stable cluster-cores and establishing statistical consistency guarantees with strong performance on real datasets and image segmentation applications.
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.