User cinch

6 votes

Accepted

Why are the generator and discriminator designed differently in this example?

answered Mar 6 at 6:58

6 votes

Accepted

Why policy improvement theorem can't be applied in case of function approximation?

answered Mar 10 at 7:28

6 votes

Can we simply remove the log term for loss in policy gradient methods?

answered Jul 29 at 0:14

5 votes

Accepted

Is it possible that Precision and Recall increase together?

answered Nov 26, 2023 at 0:59

5 votes

Accepted

Why doesn't SiLU suffer from a worse version of a "dying ReLU" problem?

answered Sep 10 at 21:33

4 votes

Accepted

Why isn’t this formula used at all?

answered Aug 1 at 8:38

4 votes

Accepted

AlphaProof and infinitary combinatorics

answered Aug 8 at 8:46

4 votes

Why is the backpropagation algorithm used to train the multilayer perceptron?

answered Apr 14, 2021 at 4:19

4 votes

Accepted

Where does the TD formula for tic-tac-toe in Sutton & Barto come from?

answered Apr 11 at 5:13

3 votes

Accepted

How does BackProp avoid bias for the last input used for training?

answered Jul 14 at 20:23

3 votes

Why exactly do we need the learning rate in gradient descent?

answered Jul 25 at 22:31

3 votes

Accepted

Likelihood function for Gaussian Discriminant Analsis

answered Feb 29 at 5:16

3 votes

What are reward networks in reinforcement learning?

answered Mar 3 at 5:59

3 votes

Why different noise in GAN generate different images?

answered Mar 13 at 3:42

3 votes

Accepted

Policiy Gradient Infinite-Horizon Analysis

answered Mar 24 at 6:58

3 votes

Accepted

Is the Bellman backup unbiased?

answered Mar 31 at 5:24

3 votes

Accepted

What is the optimal policy for this MDP?

answered Apr 7 at 5:42

3 votes

Why do activation functions in neural networks have to be non-polynomial to approximate any function?

answered Nov 22, 2022 at 6:30

3 votes

Accepted

My cross entropy loss gradient calculation is wrong according to the answer key

answered Dec 5, 2022 at 2:37

3 votes

Why can't Lucene search be used to power LLM applications?

answered Jul 13, 2023 at 21:24

3 votes

Accepted

What is the partial derivative $\frac{\partial y}{\partial x_1}$ in this neural network?

answered Apr 29, 2023 at 21:13

3 votes

In Q-learning, Am I the one who will define the way in which actions allow the agent to interact with the environment? And the interactions will vary?

answered Nov 11, 2023 at 0:23

3 votes

Accepted

What do you mean by "updating based on a training example/batch" in Gradient Descent?

answered Feb 5 at 6:49

3 votes

In Q-learning, states need to be just X and Y positions of the agent, or a state can be several other characteristics?

answered Nov 11, 2023 at 22:57

3 votes

Accepted

Unclear point in definition of advantage function in PPO

answered Sep 3 at 6:26

2 votes

Accepted

Why in loss function of DQN does the expectation depend on current state $s$?

answered Sep 1 at 7:55

2 votes

Accepted

Derivation of the relationship between state-value function and action-value function in SAC

answered Sep 8 at 4:47

2 votes

Can Q-learning be used to create new creative solutions by combining different factors and characteristics?

answered Nov 11, 2023 at 23:42

2 votes

What is the relation between Dynamic Programming and Reinforcement Learning?

answered Nov 13, 2023 at 23:22

2 votes

Can mini-batches for stochastic gradient be balanced but not representative of the training data?

answered Nov 18, 2023 at 22:36

Stack Exchange Network

183 Answers

Why are the generator and discriminator designed differently in this example?

Why policy improvement theorem can't be applied in case of function approximation?

Can we simply remove the log term for loss in policy gradient methods?

Is it possible that Precision and Recall increase together?

Why doesn't SiLU suffer from a worse version of a "dying ReLU" problem?

Why isn’t this formula used at all?

AlphaProof and infinitary combinatorics

Why is the backpropagation algorithm used to train the multilayer perceptron?

Where does the TD formula for tic-tac-toe in Sutton & Barto come from?

How does BackProp avoid bias for the last input used for training?

Why exactly do we need the learning rate in gradient descent?

Likelihood function for Gaussian Discriminant Analsis

What are reward networks in reinforcement learning?

Why different noise in GAN generate different images?

Policiy Gradient Infinite-Horizon Analysis

Is the Bellman backup unbiased?

What is the optimal policy for this MDP?

Why do activation functions in neural networks have to be non-polynomial to approximate any function?

My cross entropy loss gradient calculation is wrong according to the answer key

Why can't Lucene search be used to power LLM applications?

What is the partial derivative $\frac{\partial y}{\partial x_1}$ in this neural network?

In Q-learning, Am I the one who will define the way in which actions allow the agent to interact with the environment? And the interactions will vary?

What do you mean by "updating based on a training example/batch" in Gradient Descent?

In Q-learning, states need to be just X and Y positions of the agent, or a state can be several other characteristics?

Unclear point in definition of advantage function in PPO

Why in loss function of DQN does the expectation depend on current state $s$?

Derivation of the relationship between state-value function and action-value function in SAC

Can Q-learning be used to create new creative solutions by combining different factors and characteristics?

What is the relation between Dynamic Programming and Reinforcement Learning?

Can mini-batches for stochastic gradient be balanced but not representative of the training data?