Optimal and Efficient Dynamic Regret Algorithms for Non-Stationary Dueling Bandits

Authors: Aadirupa Saha, Shubham Gupta

ICML 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive simulations corroborate our results.
Researcher Affiliation Industry 1Microsoft Research, New York City, United States. 2IBM Research, Orsay, France.
Pseudocode Yes Algorithm 1 presents the pseudocode for DEX3.P.
Open Source Code No The paper does not provide any explicit statement or link indicating that the source code for the described methodology is open-source or publicly available.
Open Datasets No We simulate an environment where these values follow a Gaussian random walk. That is, for every t ∈ [T] and i < j, Pt+1(i, j) = Pt(i, j) + ϵt(i, j), where ϵt(i, j) ∼ N(0, 0.002). ... The initial values P1(i, j) ∼ Uniform(0, 1).
Dataset Splits No The paper describes generating synthetic data for simulations but does not specify distinct training, validation, and test dataset splits.
Hardware Specification No The paper does not specify any hardware details (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper does not specify any software dependencies with version numbers (e.g., programming languages, libraries, frameworks).
Experiment Setup Yes The values of parameters α, β, η, and γ for DEX3.P and DEX3.S were set in accordance with Theorems 3.3 and 4.1 (or 4.4 as appropriate from the context), respectively. ... We simulate an environment where these values follow a Gaussian random walk. That is, for every t ∈ [T] and i < j, Pt+1(i, j) = Pt(i, j) + ϵt(i, j), where ϵt(i, j) ∼ N(0, 0.002).