Multi-class Graph Clustering via Approximated Effective $p$-Resistance

Authors: Shota Saito, Mark Herbster

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we provide experiments comparing our approximated p-resistance clustering to other p-Laplacian based methods. ... Our experiment demonstrates that our algorithm outperforms the existing multi-class clustering using graph p-Laplacian and 2-resistance-based methods.
Researcher Affiliation Academia 1Department of Computer Science, University College London, London, United Kingdom.
Pseudocode Yes Algorithm 1 Clustering Algorithm via p-Resistance
Open Source Code Yes The implementation of our method is available at https://github.com/Shota SAITO/ approximated-presistance.
Open Datasets Yes Our experiments were performed on classification datasets, ionosphere, iris and wine from UCI repository. We also used Hopkins155 dataset (Tron & Vidal, 2007)...
Dataset Splits No The paper does not explicitly state specific training, validation, and test dataset splits by percentages, counts, or by referencing predefined splits with citations. It mentions using classification datasets for two-class and multi-class tasks but without specifying how data was partitioned for training and validation.
Hardware Specification Yes Our experiment was conducted on Mac Studio with M1 Max Processor and 32Gi B RAM.
Software Dependencies No The paper mentions: "Also, we use an Intel binary Matlab translated by Rosetta, which is a standard use in Mac OS with Apple Silicon environment." While Matlab is named, a specific version number for Matlab itself or any other libraries/solvers used is not provided.
Experiment Setup Yes We used free parameters µPt0.04, 0.06, 0.08, 0.1, 1u, σPt10 3, . . . , 102u and p Pt1.1, 1.4, . . . , 2.9, 5, 10, 100, 1000u.