Feature Clustering for Support Identification in Extreme Regions

Authors: Hamid Jalalzai, Rémi Leluc

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical experiments provide strong empirical evidence of the relevance of our approach. and We perform some numerical experiments in Section 5 to highlight the performance of our method and we finally conclude in Section 6. and We focus on popular machine learning tasks of feature clustering and anomaly detection to compare the performance of our algorithm against state-of-the-art methods for extreme events. (Section 5, Numerical Experiments).
Researcher Affiliation Academia 1Telecom Paris, Institut Polytechnique de Paris, France 2INRIA, Institut Polytechnique de Paris, France. Correspondence to: Hamid Jalalzai <hamid.jalalzai@inria.fr>, R emi Leluc <remi.leluc@telecom-paris.fr>.
Pseudocode Yes A detailed pseudo-code of MEXICO is provided below in Algorithm 1.
Open Source Code Yes For ease of reproducibility, the code is available in the supplementary material.
Open Datasets Yes Five reference AD datasets are studied: shuttle, forestcover, http, SF and SA. Table 5 in the Appendix provides further dataset details. The algorithms are trained and tested on the same datasets... We consider simulated data from an (asymmetric) logistic distribution where the dependence structure of extremes can be specified (see Appendix B.2). and references KDDCup (1999) and Lichman (2013). UCI machine learning repository.
Dataset Splits No The parameter setting is the following: dimension p {75, 100, 150, 200}, number of train samples ntrain = 1000 and test samples ntest = 100. The paper does not specify percentages or sample counts for a distinct validation set.
Hardware Specification No The paper does not specify any hardware details such as GPU/CPU models, memory, or cloud resources used for running the experiments.
Software Dependencies No We use the metrics implemented by Scikit-Learn (Pedregosa et al., 2011). No specific version numbers for Scikit-Learn or other software dependencies are provided.
Experiment Setup No More details about the preprocessing, model tuning and additional results are available in the supplementary material. (Section 5.2). However, specific hyperparameters or training settings are not provided in the main text.