Online Algorithms for Rent-Or-Buy with Expert Advice

Authors: Sreenivas Gollapudi, Debmalya Panigrahi

ICML 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test the efficacy of our algorithms via simulations. We set the buying cost b = 1. (The actual value of b is unimpor-tant because we can scale all values by b.) We choose the actual outcome x to be a value uniformly drawn from [0, 2b]. We vary the number of experts from 1 to 8 and set their associated predictions to x + ϵ where ϵ is drawn from a normal distribution of mean 0 and standard deviation σ. To verify consistency and robustness of our algorithms, we vary σ from 0 to 2. Finally, for the algorithm in Fig. 4, we consider values of 0.1, 0.5, and 0.9 for the meta parameter λ. We label the algorithm defined in Figure 1 consistent; it s extension to handle non-zero prediction errors (see Section 3) as robust; and the robust and consistent algorithm in Section 4 as hybrid. Figure 3 illustrates the relative performance of our algorithms. We make three observations.
Researcher Affiliation Collaboration 1Google Research 2Department of Computer Science, Duke University.
Pseudocode Yes Figure 1. The algorithm for k experts with zero error; Figure 2. The algorithm for k experts with non-zero error; Figure 4. The hybrid algorithm for k experts
Open Source Code No No explicit statement or link providing access to the source code for the described methodology.
Open Datasets No We test the efficacy of our algorithms via simulations. We set the buying cost b = 1. ... We choose the actual outcome x to be a value uniformly drawn from [0, 2b]. We vary the number of experts from 1 to 8 and set their associated predictions to x + ϵ where ϵ is drawn from a normal distribution of mean 0 and standard deviation σ.
Dataset Splits No The paper describes simulations where data is generated for each trial, but does not specify explicit train/validation/test splits as typically used with fixed datasets.
Hardware Specification No No specific hardware details (GPU/CPU models, memory, or specific computer specifications) are mentioned for running the experiments.
Software Dependencies No No specific software dependencies with version numbers are mentioned.
Experiment Setup Yes We set the buying cost b = 1. We choose the actual outcome x to be a value uniformly drawn from [0, 2b]. We vary the number of experts from 1 to 8 and set their associated predictions to x + ϵ where ϵ is drawn from a normal distribution of mean 0 and standard deviation σ. To verify consistency and robustness of our algorithms, we vary σ from 0 to 2. Finally, for the algorithm in Fig. 4, we consider values of 0.1, 0.5, and 0.9 for the meta parameter λ.