package obandit

  1. Overview
  2. Docs
Ocaml Multi-Armed Bandits

Install

Dune Dependency

Authors

Maintainers

Sources

obandit-0.3.4.tbz
md5=f5aa2c86eb25d4fad308d3de0dbc9288

CHANGES.md.html

v0.3.4 2018-03-16 Grenoble

  • opam url updates

  • experimental validation fixes on 0.3.

  • bugfix in exp3 reward ranges

  • added cli interface

  • added validation suite

v0.2.3.2 2017-04-13 Grenoble

  • fix delegate to use github based documentation.

v0.2.3.1 2017-04-13 Grenoble

  • fix delegate to use github based releases.

v0.2.3 2017-04-12 Grenoble

  • Change opam file to point to github issues and pages.

  • Add custom rule for building the documentation with mathjax as ocamlbuild plugin.

v0.2.1 2017-03-23 Grenoble

  • Improve the doubling trick. If the observed value is larger than twice the range, we should do more than double the window.

  • Documentation Fixes and use of Mathjax.

v0.2 2017-02-17 Grenoble

  • Interface changes. Introducing immutable versions of the algorithms.

  • Documentation overhaul with mathematical aspects.

  • Use the formalism from S.Bubeck and N.Cesa-Bianchi's survey "Regret analysis of Stochastic and Nonstochastic Multi-Armed Bandit Problems".

v0.1.42 2017-02-10 Grenoble

  • Change the doubling trick to a Range, add documentation.

v0.1.41 2017-02-08 Grenoble

  • Implement the doubling trick with restarting.

v0.1.38 2017-02-07 Grenoble

  • fix build command.

v0.1.37 2017-02-06 Grenoble

  • distrib folder change.

v0.1.36 2017-02-06 Grenoble

  • distrib url fix.

v0.1.35 2017-02-06 Grenoble

  • distrib url change.

v0.1.34 2017-02-06 Grenoble

  • opam file fixes v2.

v0.1.33 2017-02-06 Grenoble

  • opam file fixes.

v0.1.32 2017-02-06 Grenoble

  • changes to contact information.

v0.1.31 2017-02-06 Grenoble

  • Change build system.

v0.1.3 2017-02-06 Grenoble

  • Added deployment tools.

v0.1.2 2017-02-06 Grenoble

  • Documentation bugfix two.

v0.1.1 2017-02-06 Grenoble

  • Documentation bugfix.

v0.1.0 2017-02-06 Grenoble

  • First release with EXP3, UCB1, Epsilon Greedy.