Mastering Reinforcement Learning with Python by Enes Bilgin

Author:Enes Bilgin
Language: eng
Format: epub
Publisher: Packt Publishing Pvt Ltd
Published: 2020-12-16T00:00:00+00:00

where and is a hyperparameter. Therefore, GAE gives us another knob to control the bias-variance trade-off. Specifically, results in , which has high bias; and results in , which is equivalent to using the sampled reward-to-to minus the baseline, which has high bias. Any value of is a compromise between the two.

Let's close this section by noting that you can turn on or off the GAE in RLlib's actor-critic implementations using the config flag "use_gae" along with "lambda".

This concludes our discussion on actor-critic functions. Next, we'll look into a recent approach called trust-region methods, which have resulted in significant improvements over A2C and A3C.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.