Tight Rates for Bandit Control Beyond Quadratics

Authors: Y. Jennifer Sun, Zhou Lu

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

Reproducibility Variable Result LLM Response
Research Type Theoretical Our main contribution is an algorithm that achieves an O(T) optimal regret for bandit nonstochastic control with strongly-convex and smooth cost functions in the presence of adversarial perturbations, improving the previously known O(T 2/3) regret bound from (Cassel and Koren, 2020).
Researcher Affiliation Academia Y. Jennifer Sun Princeton University ys7849@princeton.edu Zhou Lu Princeton University zhoul@princeton.edu
Pseudocode Yes Algorithm 1 Improved Bandit Convex Optimization with Affine Memory; Algorithm 2 Improved Bandit Non-stochastic Control; Algorithm 3 Simple BCO-with-delay
Open Source Code No The paper does not provide concrete access to source code. It is a theoretical paper that presents algorithms.
Open Datasets No The paper is theoretical and does not conduct empirical studies using datasets.
Dataset Splits No The paper is theoretical and does not conduct empirical studies using datasets, therefore no dataset splits for validation are mentioned.
Hardware Specification No The paper is theoretical and does not describe any specific hardware used for experiments.
Software Dependencies No The paper is theoretical and does not list specific software dependencies with version numbers for experimental reproducibility.
Experiment Setup No The paper is theoretical and does not include an experimental setup with hyperparameters or system-level training settings.