Towards Better-than-2 Approximation for Constrained Correlation Clustering

Authors: Andreas Kalavas, Evangelos Kipouridis, Nithin Varma

ICML 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical Our main result is a proof that, to achieve a better-than-2 approximation for Constrained Correlation Clustering, it suffices to design an efficient solution to the Constrained Cluster LP. Finally, in Section 4 we give a full analysis for our main algorithm and how it achieves its approximation factor of (1.92 + ε).
Researcher Affiliation Academia 1Max Planck Institute for Informatics & Saarland Informatics Campus, Saarbr ucken, Germany 2Archimedes, Athena Research Center, Athens, Greece 3University of Cologne, Germany. Correspondence to: Andreas Kalavas <EMAIL>.
Pseudocode Yes Algorithm 1: Local Search; Algorithm 2: Local Search with penalties; Algorithm 3: Simple sampling-based 2-approximation; Algorithm 4: Pivoting Procedure; Algorithm 5: Main Algorithm.
Open Source Code No The paper discusses algorithms and their theoretical approximation factors but does not contain any statements about code release or links to repositories.
Open Datasets No The paper is theoretical and focuses on an approximation algorithm for Constrained Correlation Clustering. It defines the problem using abstract graphs and instances, but does not refer to any specific datasets or provide access information for them.
Dataset Splits No The paper is theoretical and does not conduct experiments on any specific dataset, therefore, there is no mention of dataset splits.
Hardware Specification No The paper is theoretical and focuses on algorithm design and analysis, without conducting any experiments. Therefore, no hardware specifications are mentioned.
Software Dependencies No The paper describes theoretical algorithms and their analysis, without discussing implementation details or specific software dependencies with version numbers.
Experiment Setup No The paper is theoretical, presenting algorithms and their approximation factors without conducting experiments. Consequently, no specific experimental setup details, such as hyperparameters or training configurations, are provided.