Epsilon decreasing gambler.
For a list of all members of this type, see EpsilonDecreasingGambler Members.
Public static (Shared in Visual Basic) members of this type are safe for multithreaded operations. Instance members are not guaranteed to be thread-safe.
The epsilon decreasing strategy is similar to the EpsilonGreedyGambler but the epsilon value decreases over time. This implementation provides a decreasing factor of
e0 / t where
e0 is a positive tuning parameter and
t the current round index.
The epsilon decreasing strategy is analysed in Finite time analysis of the multiarmed bandit problem. by Auer, Cesa-Bianchi and Fisher in Machine Learning (2002).
Assembly: Bandit (in Bandit.dll)