User Alberto

6 votes

1 answer

439 views

Estimating confidence interval of a parameter from the MLE of another parameter

asked Jan 17, 2022 at 18:15

6 votes

1 answer

70 views

If I set a Gaussian prior in the parameter space, can I use the variance as diagonal of the Hessian?

asked Dec 1, 2023 at 14:38

4 votes

1 answer

861 views

Does stacking multiple linear layer have some documented improvements?

asked Mar 12, 2023 at 21:28

4 votes

1 answer

635 views

How are the embedding split between multihead in transformers

asked Apr 8, 2023 at 10:36

4 votes

1 answer

56 views

Why do we handpick a specific loss for classification

asked Aug 28, 2022 at 14:43

3 votes

1 answer

284 views

Does the n >> p holds also for minibatches

asked Jun 21, 2022 at 20:33

3 votes

1 answer

93 views

Why in batch norm we don't restrict beta to be positive

asked Jan 29, 2023 at 17:54

2 votes

1 answer

119 views

Is TRPO just a "safe" version of off-policy policy iteration

asked Mar 17, 2023 at 0:50

2 votes

1 answer

63 views

MCMC seems very sensible to the evidence

asked Feb 29 at 22:41

2 votes

0 answers

113 views

Predict uncertainty in NN prediction [closed]

asked Jun 24, 2022 at 23:15

2 votes

0 answers

320 views

Why not using Convolutional pooling instead of MaxPooling to avoid invariance

asked Jul 28, 2022 at 15:13

2 votes

0 answers

178 views

Intuition behind log in kl distance

asked Jun 21, 2022 at 14:12

2 votes

1 answer

155 views

What happens if we follow the gradient of a softmax activation

asked Dec 19, 2022 at 12:24

1 vote

0 answers

35 views

Conditional GAN training

asked Aug 6, 2022 at 10:19

1 vote

0 answers

120 views

CycleGAN cycle loss

asked Aug 10, 2022 at 17:02

1 vote

1 answer

358 views

Stacked RNN to overcome vanishing gradient

asked Aug 17, 2022 at 23:06

1 vote

0 answers

37 views

How to merge 2 losses in a reasonable ratio

asked Aug 31, 2022 at 15:20

1 vote

0 answers

49 views

Why do we have first/every-visit in MC but not last visit [closed]

asked Nov 1, 2022 at 20:46

1 vote

1 answer

209 views

Regress linear regression coefficient with neural networks

asked Dec 9, 2022 at 15:53

1 vote

1 answer

44 views

Can I use the likelihood to estimate probability of draws

asked Feb 7, 2022 at 22:51

1 vote

1 answer

412 views

GAN artifacts on borders

asked Aug 4, 2022 at 12:38

1 vote

0 answers

9 views

Is backprop computed for every element in the minibatch and then averaged for every weight?

asked Jul 4, 2022 at 12:20

1 vote

0 answers

44 views

When is variational inference really powerful

asked Mar 18 at 15:19

1 vote

0 answers

28 views

why Conditional VAE require conditioning of the encoder

asked Mar 22 at 17:15

1 vote

3 answers

155 views

How do we get rid of $p(x|\theta)$ in MLE

asked Apr 11 at 22:11

1 vote

1 answer

21 views

Validation Accuracy higher if trained on training set than in validation set

asked Apr 18 at 15:09

1 vote

0 answers

34 views

Why is the gradient not depending on the loss?

asked Apr 14, 2023 at 17:59

1 vote

0 answers

51 views

Expectile loss to reduce dependent variables overestimation

asked Nov 27, 2023 at 22:22

1 vote

0 answers

93 views

Are rewards in reinforcement learning scale and shift invariant?

asked Jan 12, 2023 at 15:20

0 votes

0 answers

56 views

Logistic regression only cares about the $E[X|Y]$ for prediction

asked Jun 2 at 20:58

Stack Exchange Network

34 Questions

Estimating confidence interval of a parameter from the MLE of another parameter

If I set a Gaussian prior in the parameter space, can I use the variance as diagonal of the Hessian?

Does stacking multiple linear layer have some documented improvements?

How are the embedding split between multihead in transformers

Why do we handpick a specific loss for classification

Does the n >> p holds also for minibatches

Why in batch norm we don't restrict beta to be positive

Is TRPO just a "safe" version of off-policy policy iteration

MCMC seems very sensible to the evidence

Predict uncertainty in NN prediction [closed]

Why not using Convolutional pooling instead of MaxPooling to avoid invariance

Intuition behind log in kl distance

What happens if we follow the gradient of a softmax activation

Conditional GAN training

CycleGAN cycle loss

Stacked RNN to overcome vanishing gradient

How to merge 2 losses in a reasonable ratio

Why do we have first/every-visit in MC but not last visit [closed]

Regress linear regression coefficient with neural networks

Can I use the likelihood to estimate probability of draws

GAN artifacts on borders

Is backprop computed for every element in the minibatch and then averaged for every weight?

When is variational inference really powerful

why Conditional VAE require conditioning of the encoder

How do we get rid of $p(x|\theta)$ in MLE

Validation Accuracy higher if trained on training set than in validation set

Why is the gradient not depending on the loss?

Expectile loss to reduce dependent variables overestimation

Are rewards in reinforcement learning scale and shift invariant?

Logistic regression only cares about the $E[X|Y]$ for prediction