Generalization for multiclass classification with overparameterized linear models

Authors: Vignesh Subramanian, Rahul Arya, Anant Sahai

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

Reproducibility Variable Result LLM Response
Research Type Theoretical Not really relevant since there are no empirical experiments, only plots of the regions defined by the theorems which are themselves given by simple linear inequalities.
Researcher Affiliation Academia Vignesh Subramanian Department of Electrical Engineering and Computer Sciences University of California Berkeley Berkeley, CA-994720, USA vignesh.subramanian@eecs.berkeley.edu Rahul Arya Department of Electrical Engineering and Computer Sciences University of California Berkeley Berkeley, CA-994720, USA rahularya@berkeley.edu Anant Sahai Department of Electrical Engineering and Computer Sciences University of California Berkeley Berkeley, CA-994720, USA sahai@eecs.berkeley.edu
Pseudocode No The paper contains mathematical formulations, equations, and a problem setup description but does not include any explicit pseudocode or algorithm blocks.
Open Source Code No Not really relevant since there are no empirical experiments, only plots of the regions defined by the theorems which are themselves given by simple linear inequalities.
Open Datasets No The paper describes a theoretical setup with synthetic data generation (i.i.d Gaussian vectors, Assumption 4.1, Bi-level ensemble model 4.2) rather than using a publicly available or open dataset.
Dataset Splits No The paper focuses on theoretical analysis and does not describe experiments involving specific training, validation, or test dataset splits. It models the problem asymptotically.
Hardware Specification No Not applicable since everything could be done by hand.
Software Dependencies No Not applicable since everything could be done by hand.
Experiment Setup No The paper is theoretical and does not describe an empirical experiment setup with hyperparameters, training configurations, or system-level settings. It defines parameters for its theoretical model (p, q, r, t) but these are not experimental setup details.