Surrogate Regret Bounds for Polyhedral Losses

Authors: Rafael Frongillo, Bo Waggoner

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We provide two general results. The first gives a linear surrogate regret bound for any polyhedral (piecewise-linear and convex) surrogate, meaning that surrogate generalization rates translate directly to target rates. The second shows that for sufficiently non-polyhedral surrogates, the regret bound is a square root, meaning fast surrogate generalization rates translate to slow rates for the target. ... The proof of our upper bound, Theorem 1, is nonconstructive. After proving the main results (Sections 3 and 4), we make it constructive in Section 5.
Researcher Affiliation Academia Rafael Frongillo University of Colorado Boulder raf@colorado.edu Bo Waggoner University of Colorado Boulder bwag@colorado.edu
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code No This paper is theoretical and does not present methodology that would typically involve source code for release, nor does it make any statement about providing code.
Open Datasets No The paper is a theoretical work focusing on mathematical proofs and analyses of loss functions; it does not involve the use of datasets for empirical evaluation.
Dataset Splits No The paper is a theoretical work and does not describe experiments that would require training/validation/test dataset splits.
Hardware Specification No The paper is a theoretical study and does not describe experiments that would involve specific hardware. Therefore, no hardware specifications are mentioned.
Software Dependencies No The paper is a theoretical work and does not describe experiments that would involve specific software dependencies with version numbers.
Experiment Setup No The paper is a theoretical work and does not describe experiments. Therefore, no experimental setup details such as hyperparameters or training configurations are provided.