User Phizaz

7 votes

2 answers

787 views

Why don't people use projected Bellman error with deep neural networks?

asked Apr 12, 2019 at 5:02

5 votes

2 answers

370 views

Is there an alternative to the use of target network?

asked Feb 27, 2019 at 9:25

3 votes

1 answer

7k views

In DQN, updating target network every N steps or slowly update every step is better?

asked Feb 28, 2019 at 3:56

3 votes

1 answer

113 views

What is a non-starving policy in reinforcement learning?

asked Sep 11, 2019 at 4:10

3 votes

1 answer

332 views

What are the differences between stability and convergence in reinforcement learning?

asked Sep 19, 2019 at 4:49

3 votes

3 answers

760 views

Why they use KL divergence in Natural gradient?

asked Oct 30, 2019 at 4:19

2 votes

0 answers

418 views

Why overfitting is bad in DQN?

asked Apr 30, 2019 at 15:40

0 votes

1 answer

60 views

Why doesn't stability in prediction imply stability in control in off-policy reinforcement learning?

asked Sep 19, 2019 at 5:01

Stack Exchange Network

8 Questions

Why don't people use projected Bellman error with deep neural networks?

Is there an alternative to the use of target network?

In DQN, updating target network every N steps or slowly update every step is better?

What is a non-starving policy in reinforcement learning?

What are the differences between stability and convergence in reinforcement learning?

Why they use KL divergence in Natural gradient?

Why overfitting is bad in DQN?

Why doesn't stability in prediction imply stability in control in off-policy reinforcement learning?