04ReinforcementLearning3.4, Log-likelihood trick: from batch to online
From Wulfram Gerstner
views
comments
From Wulfram Gerstner
When going from batch to online, care has to be taken to arrive at the statistical weight of online sampling. The log-likelhood trick solves this issue.
Mediaspace will be updated on Saturday, March 29th. Users may experience minor access and performance restrictions.
(close this alert with x ↓↓ )