Quickshift++: Provably Good Initializations for Sample-Based Mean Shift
arXiv:1805.07909
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.
arXiv commentICML 2018. Code release: https://github.com/google/quickshift
References in corpus (8)
- Optimal rates for plug-in estimators of density level sets
- Pruning nearest neighbor cluster trees
- Quickshift++: Provably Good Initializations for Sample-Based Mean Shift
- Clusters and water flows: a novel approach to modal clustering through Morse theory
- Density Level Set Estimation on Manifolds with DBSCAN
- Feature Selection For High-Dimensional Clustering
- Adaptive Hausdorff estimation of density level sets
- Analysis of a Mode Clustering Diagram
Cited by in corpus (8)
- An Efficient One-Class SVM for Anomaly Detection in the Internet of Things
- Quickshift++: Provably Good Initializations for Sample-Based Mean Shift
- Faster DBSCAN via subsampled similarity queries
- Unsupervised Pre-Training for 3D Leaf Instance Segmentation
- Pre-Clustering Point Clouds of Crop Fields Using Scalable Methods
- Feature Extraction for Novelty Detection in Network Traffic
- MeanShift++: Extremely Fast Mode-Seeking With Applications to Segmentation and Object Tracking
- An Asymptotic Equivalence between the Mean-Shift Algorithm and the Cluster Tree