Comparator-Adaptive Convex Bandits

Authors: Dirk van der Hoeven, Ashok Cutkosky, Haipeng Luo

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical Our contribution is primarily theoretical, and we do not foresee any negative ethical or societal impact.Our results are summarized in Table 1. Regret is measured with respect to the total loss of an arbitrary point u Rd in the unconstrained setting, or an arbitrary point u W in the constrained setting with a decision space W contained in the unit ball. T is the total number of rounds, 1/c is radius of the largest ball contained by W, and ν is the self-concordant parameter. Both c and ν are bounded by O(d).
Researcher Affiliation Academia Dirk van der Hoeven Mathematical Institute Leiden University dirk@dirkvanderhoeven.com Ashok Cutkosky Boston University ashok@cutkosky.com Haipeng Luo Computer Science Department University of Southern California haipengl@usc.edu
Pseudocode Yes Algorithm 1 Black-Box Reduction with Full Information, Algorithm 2 Black-Box Reduction for Linear Bandits, Algorithm 5 Black-Box Comparator-Adaptive Convex Bandit Algorithm
Open Source Code No The paper does not contain any statement or link indicating that source code for the described methodology is publicly available.
Open Datasets No The paper focuses on theoretical algorithm development and regret analysis, not on empirical evaluation with datasets. Therefore, it does not mention publicly available datasets for training.
Dataset Splits No The paper is theoretical and does not involve empirical experiments with dataset splits for training, validation, or testing.
Hardware Specification No The paper is theoretical and focuses on algorithm design and regret analysis, not on empirical experiments. Therefore, it does not provide hardware specifications.
Software Dependencies No The paper is theoretical and does not mention specific software dependencies with version numbers required for replication.
Experiment Setup No The paper is theoretical and focuses on algorithm design and regret analysis, not on empirical experiments. Therefore, it does not provide details about an experimental setup, hyperparameters, or training configurations.