Random Gegenbauer Features for Scalable Kernel Methods

Authors: Insu Han, Amir Zandieh, Haim Avron

ICML 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Empirical results show that our proposed features outperform recent kernel approximation methods. and 6. Experiments section.
Researcher Affiliation Academia 1Yale University 2Max-Planck-Institut f ur Informatik 3Tel Aviv University.
Pseudocode No The paper does not include a clearly labeled pseudocode or algorithm block.
Open Source Code No The paper does not provide an explicit statement about releasing source code or a link to a code repository for the described methodology.
Open Datasets Yes For kernel ridge regression, we use 4 real-world datasets, e.g., Earth Elevation2, CO2 3, Climate4 and Protein5. 2https://github.com/fatiando/rockhound, 3https://db.cger.nies.go.jp/dataset/ODIAC/, 4http://berkeleyearth.lbl.gov/, 5https://archive.ics.uci.edu/. For kernel k-means clustering, we use 6 UCI classification datasets6. 6http://persoal.citius.usc.es/manuel.fernandez.delgado/papers/jmlr/data.tar.gz
Dataset Splits Yes For all datasets, we randomly split 90% training and 10% testing, and find the ridge parameter via the 2-fold cross-validation on the training set.
Hardware Specification No The paper does not specify the exact hardware (e.g., CPU, GPU models, or cloud instance types) used for running the experiments.
Software Dependencies No The paper mentions software such as 'scipy.integrate.quad' and 'scikit-learn' but does not provide specific version numbers for these dependencies.
Experiment Setup Yes For all kernel approximation methods, we set the final feature dimension to m = 1,024. and The number of clusters is set to the number of classes of each dataset and the number of features are commonly set to m = 512. and find the ridge parameter via the 2-fold cross-validation on the training set.