Dominant-Set Clustering Algorithm Implementation
- Login to Download
- 1 Credits
Resource Overview
Implementation of the relatively new Dominant-Set clustering algorithm, including both core clustering code and test scripts. This graph theory-based approach offers significantly lower computational complexity compared to Normalized Cut while automatically determining optimal cluster numbers.
Detailed Documentation
This resource provides an implementation of the Dominant-Set clustering algorithm, featuring both the core clustering module and comprehensive test code. The algorithm's primary distinction lies in its graph-theoretic foundation, which achieves substantially lower computational complexity than traditional methods like Normalized Cut. The implementation includes adjacency matrix construction and pairwise similarity computation modules that efficiently handle graph representation. A key advantage is its built-in cluster number determination capability, eliminating the need for manual parameter tuning through iterative optimization processes that identify dominant sets using payoff functions and replicator dynamics. The test suite validates performance on various datasets, demonstrating automatic threshold detection and cluster validation metrics.
- Login to Download
- 1 Credits