Ball Tree
Properties
tags
cscs/databases
created
29.10.2024, 14:17
modified
09.08.2025, 11:27
published
Empty
sources
Empty
topics
Ball Tree, Tree-based Indexes, Partition-based Indexes
authors
Empty
ai-assisted
No
- Partitioning: Uses hyperspheres (balls) to partition data, suited for clusters.
- Space Coverage: No, only covers regions where data exists; does not fill the entire space.
- Region Overlap: No, uses non-overlapping hyperspheres, although some empty space may remain.
- Complexity: O(log n) for balanced trees, with better performance than k-d Trees in high dimensions.
- Query Type: Suited for nearest neighbor and density-based queries in high dimensions.
- Use Case: Often used in machine learning, text/image retrieval, and high-dimensional clustering tasks.