Reproduce “On the convergence of the MLE as an estimator of the learning rate in the Exp3 algorithm” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Julien Aubert, Luc Lehéricy, Patricia Reynaud-Bouret
The authors analyze maximum-likelihood recovery of learning rates in sequential Exp3 choice models. In a two-arm fixed-loss setting, they show that some distinct rates remain difficult to distinguish as samples grow. For decreasing rates, truncated fitting bounds prediction error generally, but parameter error only in a two-arm zero-loss case; one simulated candidate probability becomes computer-zero after 20 trials.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.