Abstract
Given a graph with positive and negative edge labels, the correlation clustering problem aims to cluster the nodes so to minimize the total number of between-cluster positive and within-cluster negative edges. This problem has many applications in data mining, particularly in unsupervised learning. Inspired by the prevalence of large graphs and constantly changing data in modern applications, we study correlation clustering in dynamic, parallel (MPC), and local computation (LCA) settings. We design an approach that improves state-of-the-art runtime complexities in all these settings. In particular, we provide the first fully dynamic algorithm that runs in an expected amortized constant time, without any dependence on the graph size. Moreover, our algorithm essentially matches the approximation guarantee of the celebrated PIVOT algorithm.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 9931-9952 |
| Number of pages | 22 |
| Journal | Proceedings of Machine Learning Research |
| Volume | 235 |
| State | Published - 2024 |
| Event | 41st International Conference on Machine Learning, ICML 2024 - Vienna, Austria Duration: Jul 21 2024 → Jul 27 2024 |
Funding
We are grateful to anonymous reviewers for their valuable feedback and for suggesting to discuss the implementation of our algorithm in PRAM. K.Makarychev was supported by the NSF Awards CCF-1955351 and EECS-2216970. S. Mitrovi\u0107 was supported by the Google Research Scholar and NSF Faculty Early Career Development Program #2340048.
ASJC Scopus subject areas
- Artificial Intelligence
- Software
- Control and Systems Engineering
- Statistics and Probability