Bandit Online Linear Optimization with Hints and Queries

Authors: Aditya Bhaskara, Ashok Cutkosky, Ravi Kumar, Manish Purohit

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we include an experimental evaluation of Algorithm 1 on synthetic data.
Researcher Affiliation Collaboration 1University of Utah, Salt Lake City, UT, USA 2Boston University, Boston, MA, USA 3Google Research, Mountain View, CA, USA.
Pseudocode Yes Algorithm 1 Bandit OLO with Queries. Algorithm 2 Bandit OLO with Response Feedback. Algorithm 3 Hint Weight Learner
Open Source Code No The paper does not provide any explicit statements about making the source code available, nor does it include links to a code repository or mention code in supplementary materials.
Open Datasets No The paper states that data is generated synthetically for experiments: "For each time step t independently, the cost vector ct is generated as follows: the first coordinate of ct is fixed to be 0.5, and the remaining d 1 coordinates are drawn uniformly at random from a (d 1)-dimensional sphere of radius 1 0.52 so that each cost vector has unit length." and "ct = p (1, 0, 0) + (1 p) ut where ut is a uniformly random unit vector on the sphere in R3."
Dataset Splits No The paper describes synthetic data generation for online learning experiments, but it does not specify explicit training/validation/test dataset splits.
Hardware Specification No The paper describes the experimental setup and data generation but does not provide any specific details regarding the hardware (e.g., GPU/CPU models, memory) used to run the experiments.
Software Dependencies No The paper describes the experimental setup but does not specify any software dependencies or their version numbers (e.g., specific libraries, frameworks, or solvers with version information).
Experiment Setup Yes For each time step t independently, the cost vector ct is generated as follows: the first coordinate of ct is fixed to be 0.5, and the remaining d 1 coordinates are drawn uniformly at random from a (d 1)-dimensional sphere of radius 1 0.52 so that each cost vector has unit length. We set B = 0, i.e., there are no bad query responses and set the time horizon T = 5000.