Naive imputation implicitly regularizes high-dimensional linear models

Authors: Alexis Ayme, Claire Boyer, Aymeric Dieuleveut, Erwan Scornet

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments illustrate our findings.
Researcher Affiliation Academia 1Sorbonne Universit e, CNRS, Laboratoire de Probabilit es, Statistique et Mod elisation (LPSM), F-75005 Paris, France 2CMAP, UMR7641, Ecole Polytechnique, IP Paris, 91128 Palaiseau, France.
Pseudocode No Section 4.1 describes the SGD algorithm step-by-step in prose but does not include a clearly labeled pseudocode or algorithm block.
Open Source Code No The paper does not provide any links or explicit statements about releasing open-source code for the described methodology.
Open Datasets No The paper states 'We generate n = 500 complete input data according to a normal distribution with two different covariance structures.' indicating simulated data, not a publicly available dataset.
Dataset Splits No The paper describes data simulation and evaluation on test samples but does not specify train/validation/test splits, percentages, or cross-validation methodology.
Hardware Specification No The paper mentions that regressors are 'implemented in scikit-learn', but it does not specify any hardware details such as CPU, GPU models, or memory.
Software Dependencies No The paper mentions 'implemented in scikit-learn (Pedregosa et al., 2011)' but does not provide specific version numbers for scikit-learn or any other software dependencies.
Experiment Setup Yes Under Assumption 4, choosing a constant learning rate γ = 1 κTr(Σ) n leads to... and with starting point θ0 = 0 and learning rate γ = 1 dκL2 n, satisfies...