I don’t understand this part… I would say the advantage is just a whitened rewards, which is what…

Since there is no negative reward, r(a3) and r(a1) are both greater or equal to 0, implying that Q*(s, a3) and Q*(s, a2) are not very different. Thus, the agent will only have little preference of a2 over a3 in this situation.

Introduction to Various Reinforcement Learning Algorithms. Part II (TRPO, PPO)

1.8K

Kung-Hsiang, Huang (Steeve)

Fangda Han

1 min readJul 13, 2018

I don’t understand this part… I would say the advantage is just a whitened rewards, which is what neural networks would like to have as inputs

Sign up to discover human stories that deepen your understanding of the world.

Free

Distraction-free reading. No ads.

Organize your knowledge with lists and highlights.

Tell your story. Find your audience.

Membership

Read member-only stories

Support writers you read most

Earn money for your writing

Listen to audio narrations

Read offline with the Medium app

Written by Fangda Han

22 Followers

32 Following

No responses yet

Write a response

What are your thoughts?

Also publish to my profile

Help
Status
About
Careers
Press
Blog
Privacy
Terms
Text to speech
Teams