Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1]

Robust high dimensional learning for Lipschitz and convex losses

Authors: Chinot Geoffrey, Lecué Guillaume, Lerasle Matthieu

JMLR 2020 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental A short simulation study illustrating our theoretical findings is presented in Section 7. We illustrate this principle in a Simulations section where a minmax MOM version of classical proximal descent algorithms are turned into robust to outliers algorithms.
Researcher Affiliation Academia Chinot Geoffrey EMAIL Department of Statistics ETH Zurich... Lecu e Guillaume EMAIL Department of Statistics ENSAE CREST... Lerasle Matthieu EMAIL Department of Statistics ENSAE CREST
Pseudocode Yes Algorithm 1: Proximal Descent-Ascent gradient method with median blocks. Algorithm 2: Proximal gradient descent algorithm.
Open Source Code No The paper does not contain any explicit statement about releasing source code or provide a link to a code repository. It describes algorithms and methods but does not offer concrete access to its implementation.
Open Datasets No First framework: X is a standard Gaussian random vector in Rp and ζ is a real-valued standard Gaussian variable independent of X with variance σ2. This describes synthetic data generation, not the use of a publicly available dataset.
Dataset Splits No The error rate is the proportion of misclassification on a test dataset. However, no specific details like percentages, sample counts, or methodology for the splits are given.
Hardware Specification No The paper does not mention any specific hardware used for running the simulations. The 'Simulations' section (Section 7) describes the algorithms and data generation but omits hardware specifications.
Software Dependencies No The paper does not explicitly state any specific software or library names with version numbers used for the implementation or experiments.
Experiment Setup No The number of blocks K is chosen by MOM cross-validation (see Lecu e and Lerasle (2017b) for more precision on that procedure). The sequences of stepsizes are constant along the algorithm (ηt)t := η and ( ηt)t = η and are also chosen by MOM cross-validation. While N, p, s are given (e.g., N = 1000, p = 400 and s = 30), specific resulting values for hyperparameters like λ, K, or η are not provided.