User Neil Slater

211 votes

Accepted

What is the "dying ReLU" problem in neural networks?

answered May 9, 2015 at 14:25

99 votes

Accepted

Cross-entropy loss explanation

answered Jul 10, 2017 at 11:56

70 votes

Accepted

Should we apply normalization to test data as well?

answered Feb 8, 2018 at 17:04

60 votes

What is the difference between "equivariant to translation" and "invariant to translation"

answered Jan 4, 2017 at 20:39

55 votes

Accepted

What is "experience replay" and what are its benefits?

answered Jul 19, 2017 at 7:25

47 votes

Accepted

What exactly is bootstrapping in reinforcement learning?

answered Jan 23, 2018 at 8:22

43 votes

Accepted

What is Ground Truth

answered Mar 24, 2017 at 12:17

39 votes

Accepted

What is the best Keras model for multi-class classification?

answered Feb 1, 2016 at 16:41

37 votes

Accepted

How to decide neural network architecture?

answered Jul 7, 2017 at 6:33

32 votes

The cross-entropy error function in neural networks

answered Dec 10, 2015 at 16:10

31 votes

Accepted

Strange behavior with Adam optimizer when training for too long

answered Nov 22, 2017 at 21:54

26 votes

Pandas Dataframe to DMatrix

answered Jul 15, 2016 at 14:12

25 votes

Accepted

Why are variables of train and test data defined using the capital letter (in Python)?

answered Mar 15, 2017 at 7:49

25 votes

Accepted

Are there free cloud services to train machine learning models?

answered Nov 3, 2017 at 13:17

25 votes

Accepted

Deep Neural Network - Backpropogation with ReLU

answered May 30, 2017 at 15:51

25 votes

Accepted

Should one hot vectors be scaled with numerical attributes

answered May 14, 2018 at 18:11

19 votes

Accepted

Can Reinforcement learning be applied for time series forecasting?

answered Aug 30, 2018 at 9:06

19 votes

Accepted

How to deal with string labels in multi-class classification with keras?

answered Mar 11, 2017 at 14:23

18 votes

Why use both validation set and test set?

answered Apr 14, 2017 at 6:43

18 votes

Accepted

Extra output layer in a neural network (Decimal to binary)

answered Jul 31, 2015 at 9:26

17 votes

Why convolute if Max Pooling is just going to downsample the image anyway?

answered Sep 21, 2016 at 11:00

17 votes

Accepted

Which functions neural net can't approximate

answered Apr 6, 2017 at 7:22

17 votes

Accepted

Why TensorFlow can't fit simple linear model if I am minimizing absolute mean error instead of the mean squared error?

answered Nov 17, 2016 at 13:31

17 votes

Could Deep Learning be used to crack encryption?

answered Jan 31, 2017 at 7:57

17 votes

Accepted

Python implementation of cost function in logistic regression: why dot multiplication in one expression but element-wise multiplication in another

answered Aug 22, 2017 at 10:48

17 votes

Accepted

RL Advantage function why A = Q-V instead of A=V-Q?

answered Sep 1, 2018 at 8:40

16 votes

What is a 1D Convolutional Layer in Deep Learning?

answered Feb 28, 2017 at 10:03

14 votes

What is weight and bias in deep learning?

answered May 21, 2017 at 7:16

14 votes

Accepted

Why my training and validation loss is not changing?

answered Jun 9, 2017 at 6:58

14 votes

Accepted

Is Neural Style Transfer considered supervised or unsupervised learning?

answered Nov 13, 2017 at 23:25

Stack Exchange Network

443 Answers

What is the "dying ReLU" problem in neural networks?

Cross-entropy loss explanation

Should we apply normalization to test data as well?

What is the difference between "equivariant to translation" and "invariant to translation"

What is "experience replay" and what are its benefits?

What exactly is bootstrapping in reinforcement learning?

What is Ground Truth

What is the best Keras model for multi-class classification?

How to decide neural network architecture?

The cross-entropy error function in neural networks

Strange behavior with Adam optimizer when training for too long

Pandas Dataframe to DMatrix

Why are variables of train and test data defined using the capital letter (in Python)?

Are there free cloud services to train machine learning models?

Deep Neural Network - Backpropogation with ReLU

Should one hot vectors be scaled with numerical attributes

Can Reinforcement learning be applied for time series forecasting?

How to deal with string labels in multi-class classification with keras?

Why use both validation set and test set?

Extra output layer in a neural network (Decimal to binary)

Why convolute if Max Pooling is just going to downsample the image anyway?

Which functions neural net can't approximate

Why TensorFlow can't fit simple linear model if I am minimizing absolute mean error instead of the mean squared error?

Could Deep Learning be used to crack encryption?

Python implementation of cost function in logistic regression: why dot multiplication in one expression but element-wise multiplication in another

RL Advantage function why A = Q-V instead of A=V-Q?

What is a 1D Convolutional Layer in Deep Learning?

What is weight and bias in deep learning?

Why my training and validation loss is not changing?

Is Neural Style Transfer considered supervised or unsupervised learning?