On Approximate Thompson Sampling with Langevin Algorithms

Authors: Eric Mazumdar, Aldo Pacchiano, Yian Ma, Michael Jordan, Peter Bartlett

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

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we conclude in Section F by validating these theoretical results in numerical simulations where we find that Thompson sampling with our approximate sampling schemes maintain the desirable performance of exact Thompson sampling.
Researcher Affiliation Collaboration 1Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, USA 2Google Research 3Halıcıo glu Data Science Institute, University of California, San Diego, USA 4Department of Electrical Engineering and Computer Sciences, University of California, Berkeley, USA.
Pseudocode Yes Algorithm 1 Thompson sampling; Algorithm 2 (Stochastic Gradient) Langevin Algorithm for Arm a
Open Source Code No The paper does not provide any statements about open-sourcing code, nor does it provide a link to a code repository.
Open Datasets No The paper mentions "numerical simulations" but does not provide details about specific publicly available datasets, nor does it include links or citations for dataset access.
Dataset Splits No The paper does not provide specific details about training, validation, or test dataset splits (e.g., percentages, sample counts, or cross-validation methods).
Hardware Specification No The paper mentions running "numerical simulations" but does not specify any hardware details such as GPU/CPU models, memory, or cloud computing resources used.
Software Dependencies No The paper does not specify any software dependencies with version numbers.
Experiment Setup Yes If we take the batch size k = O κ2 a , step size h(n) = O 1 n 1 κa La and number of steps N = O κ2 a in the SGLD algorithm