Online Prediction with Selfish Experts

Authors: Tim Roughgarden, Okke Schrijvers

NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, in Section 6 we show simulations that indicate that different IC methods show similar regret behavior, and that their regret is substantially better than that of the non-IC standard algorithms, suggesting that the worst-case characterization we prove holds more generally. (Section 6: Simulations, Figure 1a, Figure 1b, Table 1)
Researcher Affiliation Academia Tim Roughgarden Department of Computer Science Stanford University Stanford, CA 94305 tim@cs.stanford.edu Okke Schrijvers Department of Computer Science Stanford University Stanford, CA 94305 okkes@cs.stanford.edu
Pseudocode No The paper describes algorithms (Weighted Majority (WM) and Randomized Weighted Majority (RWM)) and their update rules verbally and mathematically, but it does not present them in a formal pseudocode block or algorithm box.
Open Source Code No The paper does not provide an explicit statement about releasing source code or a link to a code repository for the methodology described.
Open Datasets No The paper mentions a "simple two-state hidden Markov model (HMM)" as a data-generating process for simulations but does not provide any access information (link, citation, repository) for the specific dataset generated or used in their simulations.
Dataset Splits No The paper describes simulations using a data-generating process but does not specify any training, validation, or test dataset splits for these simulations.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the simulations (e.g., GPU/CPU models, memory, cloud resources).
Software Dependencies No The paper does not mention any specific software dependencies or their version numbers used in the simulations.
Experiment Setup Yes For the IC methods, experts report p(t)i , for the standard algorithm p(t)i = 1 if b(t)i = 0 otherwise. The y axis is the ratio of the total loss of each of the algorithms to the performance of the best expert at that time. The plot is for 10 experts, T = 10, 000, = 10 2, and the randomized versions of the algorithms, averaged over 30 runs. Varying model parameters and the deterministic version show similar results.