how does the policy gradient theorem reduce the variance?
http://adv-ml-2017.wikidot.com/forum/t-2315060/how-does-the-policy-gradient-theorem-reduce-the-variance
Posts in the discussion thread "how does the policy gradient theorem reduce the variance?"Sat, 17 Apr 2021 05:52:37 +0000http://adv-ml-2017.wikidot.com/forum/t-2315060#post-2848373how does the policy gradient theorem reduce the variance?
http://adv-ml-2017.wikidot.com/forum/t-2315060/how-does-the-policy-gradient-theorem-reduce-the-variance#post-2848373
Sat, 10 Jun 2017 08:37:39 +0000Ido Hadanny3026643
I think I understand the derivation of the policy gradient theorem presented in the scribe rl_class1, section 2.3. However, I do not understand why does it reduce the variance? Is it because the new expression for the gradient has less terms?
]]>