Learning the Learning Rate for Prediction with Expert Advice

Authors: Wouter M. Koolen, Tim van Erven, Peter Grünwald

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In Appendix A we describe the data used to generate Figure 1 and explain why the regret obtained by LLR is significantly smaller than the regret of Ada Hedge, FTL and all other tunings described above.
Researcher Affiliation Academia Wouter M. Koolen Queensland University of Technology and UC Berkeley wouter.koolen@qut.edu.au Tim van Erven Leiden University, the Netherlands tim@timvanerven.nl Peter D. Gr unwald Leiden University and Centrum Wiskunde & Informatica, the Netherlands pdg@cwi.nl
Pseudocode Yes Algorithm 1 LLR(πah, π ). The grid η1, η2, . . . and weights π1, π2, . . . are defined in (8) and (12).
Open Source Code No The paper does not provide any links to open-source code or explicit statements about code availability.
Open Datasets No The paper uses 'Example data (details in Appendix A)' which describes how data was generated for illustrative purposes, but it does not refer to a publicly available or open dataset with access information.
Dataset Splits No The paper describes generating example data for Figure 1, but does not mention specific training/validation/test dataset splits.
Hardware Specification No The paper does not provide any specific hardware details used for running experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers.
Experiment Setup No The paper describes the LLR algorithm itself, but it does not specify concrete hyperparameter values or system-level training settings for an experimental setup.