Implements a few online algorithms in the setting of online learning with expert advice:

• External regret minimization: exponentially-weighted average forecaster
• Internal regret minimization: reduction to external regret minimization from Cesa-Bianchi and Lugosi (chapter 4)
• Online calibration: via internal regret minimization
• Online recalibration: uses online calibration method as a subroutine

See ./examples folder for details on how to run the algorithms.