Trading-Off Payments and Accuracy in Online Classification with Paid Stochastic Experts

Authors: Dirk Van Der Hoeven, Ciara Pike-Burke, Hao Qiu, Nicolò Cesa-Bianchi

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

Reproducibility Variable Result LLM Response
Research Type Experimental Our algorithm is empirically evaluated on synthetic data. ... Our experiments use two sets of productivity functions defined on a uniform (random) grid C of N payments on [1/2, 1].
Researcher Affiliation Academia 1Korteweg-de Vries Institute for Mathematics University of Amsterdam, Amsterdam, The Netherlands 2Department of Mathematics, Imperial College London, London, UK 3Universit a degli Studi di Milano, Milan, Italy 4Politecnico di Milano, Milan, Italy.
Pseudocode Yes Algorithm 1 LCB-GAPTRON
Open Source Code No The paper does not contain any explicit statement or link indicating that the source code for the described methodology is publicly available.
Open Datasets No Our algorithm is empirically evaluated on synthetic data. ... We sampled N numbers c1, . . . , c N uniformly at random from [1/2, 1] and defined pj(ci) = ci for all i [N] and all j [K].
Dataset Splits No The paper uses synthetic data for experiments but does not explicitly describe train, validation, or test dataset splits with percentages or sample counts.
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 list specific software dependencies with version numbers used for the experiments.
Experiment Setup Yes Figure 1. Cumulative cost over time for the choice of parameters K = 5, N = 5, T = 10^4, λ = 10^-2. ... Figure 2. Total cost for the following choices of parameters: K = 10, N = 10, T = 10^5, λ = 10^-2 (top plot), K = 10, N = 10, T = 10^5, λ = 10^-3 (central plot), K = 20, N = 50, T = 10^5, λ = 10^-2 (bottom plot). ... The error bars show the standard deviation of the cost averaged over 20 repetitions.