On the Efficiency of ERM in Feature Learning

Authors: Ayoub El Hanchi, Chris J. Maddison, Murat A. Erdogdu

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We study the performance of empirical risk minimization (ERM) on regression tasks with square loss and over model classes induced by arbitrary collections of features maps. Our first main result is an asymptotic characterization of the quantiles of the excess risk of any sequence of empirical risk minimizers in our setting, which vastly generalizes that of Theorem 1.
Researcher Affiliation Academia Ayoub El Hanchi University of Toronto & Vector Institute aelhan@cs.toronto.edu Chris J. Maddison University of Toronto & Vector Institute cmaddis@cs.toronto.edu Murat A. Erdogdu University of Toronto & Vector Institute erdogdu@cs.toronto.edu
Pseudocode No The paper does not include any sections or figures explicitly labeled "Pseudocode" or "Algorithm", nor are there any structured algorithmic steps presented.
Open Source Code No The paper states in the NeurIPS checklist, "The paper does not include experiments." and "The answer NA means that the paper does not include experiments requiring code." There is no mention or link to source code.
Open Datasets No The NeurIPS checklist states: "The paper does not include experiments." and "The answer NA means that the paper does not include experiments." There is no mention of datasets used for training.
Dataset Splits No The NeurIPS checklist states: "The paper does not include experiments." and "The answer NA means that the paper does not include experiments." There is no mention of validation splits.
Hardware Specification No The NeurIPS checklist states: "The paper does not include experiments." and "The answer NA means that the paper does not include experiments." There is no mention of hardware specifications.
Software Dependencies No The NeurIPS checklist states: "The paper does not include experiments." and "The answer NA means that the paper does not include experiments." There is no mention of specific software dependencies with version numbers.
Experiment Setup No The NeurIPS checklist states: "The paper does not include experiments." and "The answer NA means that the paper does not include experiments." There is no mention of experimental setup details like hyperparameters.