Fast Rates for Bandit PAC Multiclass Classification

Authors: Liad Erez, Alon Peled-Cohen, Tomer Koren, Yishay Mansour, Shay Moran

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

Reproducibility Variable Result LLM Response
Research Type Theoretical Our main contribution is in designing a novel learning algorithm for the agnostic (e, δ)-PAC version of the problem, with sample complexity of O (poly(K) + 1/e2) log(|H|/δ) for any finite hypothesis class H.
Researcher Affiliation Collaboration Liad Erez Tel-Aviv University liaderez@mail.tau.ac.il Alon Cohen Tel-Aviv University Google Research alonco@tauex.tau.ac.il Tomer Koren Tel-Aviv University Google Research tkoren@tauex.tau.ac.il Yishay Mansour Tel-Aviv University Google Research mansour.yishay@gmail.com Shay Moran Technion Google Research shaymoran1@gmail.com
Pseudocode Yes Algorithm 1 Bandit PAC Multiclass Classification via Log Barrier Stochastic Optimization, Algorithm 2 Stochastic Frank-Wolfe with SPIDER gradient estimates
Open Source Code No The paper does not provide an explicit statement about releasing code or a link to a code repository. The NeurIPS checklist indicates it is a theoretical paper.
Open Datasets No The paper is theoretical and does not use external public datasets for training. It describes an internal process of constructing a dataset 'S' within the algorithm, but this is not a publicly available dataset in the typical sense.
Dataset Splits No The paper is theoretical and does not describe training, validation, or test dataset splits for empirical experiments.
Hardware Specification No The paper is theoretical and does not describe specific hardware used to run experiments.
Software Dependencies No The paper is theoretical and does not specify software dependencies with version numbers for reproducibility.
Experiment Setup No The paper is theoretical and does not provide details about an empirical experimental setup, such as hyperparameters or training configurations.