Cluster Explanation via Polyhedral Descriptions

Authors: Connor Lawless, Oktay Gunluk

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

Reproducibility Variable Result LLM Response
Research Type Experimental We present numerical experiments on a number of real world clustering datasets and show that our approach performs favorably compared to state-of-the-art cluster description approaches.
Researcher Affiliation Academia 1Operations Research and Information Engineering, Cornell University, USA. Correspondence to: Connor Lawless <cal379@cornell.edu>.
Pseudocode No The paper describes mathematical formulations and procedures but does not include structured pseudocode blocks or sections explicitly labeled "Algorithm".
Open Source Code No The paper does not provide an explicit statement about the release of source code for the methodology or a link to a code repository.
Open Datasets Yes To evaluate our approach we ran experiments on a suite of clustering datasets from the UCI Machine Learning repository (Asuncion & Newman, 2007).
Dataset Splits No The paper does not explicitly specify train/validation/test dataset splits by percentages, counts, or by referencing predefined splits with citations for reproducibility.
Hardware Specification Yes All models were implemented in python using Gurobi 9.1 and run on a computer with 16 GB of RAM and a 2.7 GHz processor.
Software Dependencies Yes All models were implemented in python using Gurobi 9.1
Experiment Setup Yes For the following experiments we used κ = 0.05. ... For all results we set a 300 second time limit on the overall column generation procedure and a 30 second time limit on solving an individual pricing problem. ... To construct an initial set of candidate half-spaces, for each cluster we enumerate the p maximum and minimum values for each feature (p = 10 for the following experiments)... We perform the grouping by using hierarchical clustering with a maximum linkage of χ = 0.05.