User volperossa

530 votes

23 answers

286k views

Why square the difference instead of taking the absolute value in standard deviation?

c4il

5,545

asked Jul 19, 2010 at 21:04

372 votes

7 answers

1.5m views

How to normalize data to 0-1 range?

normalization

Angelo

4,101

asked Sep 23, 2013 at 15:18

347 votes

5 answers

371k views

What is the trade-off between batch size and number of iterations to train a neural network?

neural-networks train

Franck Dernoncourt

43.3k

asked Aug 5, 2015 at 21:19

239 votes

4 answers

108k views

ROC vs precision-and-recall curves

machine-learning roc precision-recall

Amelio Vazquez-Reina

18k

asked Feb 14, 2011 at 17:10

144 votes

8 answers

88k views

Why L1 norm for sparse models

regression lasso regularization ridge-regression

Yongwei Xing

1,623

asked Dec 11, 2012 at 7:25

94 votes

8 answers

126k views

How to compute precision/recall for multiclass-multilabel classification?

machine-learning classification precision-recall multi-class

Vam

1,265

asked Jan 23, 2012 at 12:54

87 votes

6 answers

58k views

Why is the L2 regularization equivalent to Gaussian prior?

regression references regularization

Anonymous

1,179

asked Jul 27, 2015 at 14:59

81 votes

2 answers

144k views

How to normalize data between -1 and 1?

dataset normalization

covfefe

1,129

asked Oct 26, 2015 at 1:02

78 votes

6 answers

109k views

What are good initial weights in a neural network?

neural-networks normalization

elmes

907

asked Jan 12, 2013 at 21:26

77 votes

2 answers

69k views

What is the difference between ZCA whitening and PCA whitening?

pca dimensionality-reduction image-processing

RockTheStar

11.7k

asked Oct 1, 2014 at 7:22

74 votes

1 answer

60k views

What are the shortcomings of the Mean Absolute Percentage Error (MAPE)?

accuracy mape

Stephan Kolassa

99.4k

asked Aug 25, 2017 at 8:49

64 votes

3 answers

61k views

Clustering with K-Means and EM: how are they related?

machine-learning clustering data-mining k-means expectation-maximization

Myna

793

asked Nov 18, 2013 at 11:47

61 votes

5 answers

139k views

Backpropagation with Softmax / Cross Entropy

backpropagation derivative softmax cross-entropy

micha

713

asked Sep 17, 2016 at 23:32

60 votes

4 answers

77k views

What is the definition of a "feature map" (aka "activation map") in a convolutional neural network?

neural-networks deep-learning conv-neural-network

Atlas7

703

asked Jul 16, 2017 at 14:16

59 votes

5 answers

125k views

Which loss function is correct for logistic regression?

logistic loss-functions

xtt

724

asked Dec 11, 2016 at 17:05

48 votes

6 answers

49k views

Why zero correlation does not necessarily imply independence

correlation independence

Victor

6,055

asked Oct 31, 2015 at 2:33

46 votes

4 answers

19k views

What are the factors that cause the posterior distributions to be intractable?

bayesian approximation inference

Nick

3,377

asked Nov 11, 2010 at 0:33

43 votes

5 answers

48k views

Entropy of an image

spatial entropy image-processing

Davor Josipovic

1,020

asked Sep 16, 2016 at 6:35

42 votes

4 answers

14k views

How can SVM 'find' an infinite feature space where linear separation is always possible?

svm feature-selection kernel-trick

user36162

561

asked Dec 23, 2013 at 11:51

33 votes

6 answers

28k views

What is difference between 'transfer learning' and 'domain adaptation'?

deep-learning terminology conv-neural-network transfer-learning domain-adaptation

mrgloom

1,817

asked Feb 6, 2017 at 16:01

31 votes

1 answer

7k views

Who invented k-fold cross-validation?

cross-validation references history

Dikran Marsupial

49.3k

asked Apr 18, 2012 at 16:48

27 votes

1 answer

7k views

Backpropagation on a convolutional layer

machine-learning neural-networks conv-neural-network backpropagation

Edv Beq

531

asked Feb 2, 2018 at 1:11

22 votes

3 answers

7k views

Backpropagation algorithm and error in hidden layer

machine-learning neural-networks backpropagation

HIGGINS

489

asked Dec 10, 2010 at 19:21

22 votes

3 answers

28k views

Why do we divide by the standard deviation and not some other standardizing factor before doing PCA?

machine-learning pca mathematical-statistics

Charlie Parker

6,096

asked Jan 19, 2015 at 23:24

21 votes

4 answers

51k views

Comparing two histograms using Chi-Square distance

chi-squared-test distance image-processing histogram correspondence-analysis

Anton Holovin

313

asked Nov 29, 2015 at 12:34

19 votes

3 answers

13k views

How to do data augmentation and train-validate split?

machine-learning classification cross-validation dataset data-augmentation

yangjie

293

asked Oct 5, 2015 at 10:43

17 votes

4 answers

35k views

In CNN, are upsampling and transpose convolution the same?

machine-learning neural-networks conv-neural-network transposed-convolution

RockTheStar

11.7k

asked Dec 22, 2016 at 0:38

15 votes

1 answer

8k views

Softmax overflow [closed]

softmax numerics

Dlmss

163

asked Sep 24, 2017 at 20:16

14 votes

3 answers

24k views

What is the difference between linearly dependent and linearly correlated?

correlation non-independent

Happy Mittal

243

asked Jun 28, 2012 at 9:02

13 votes

5 answers

5k views

Intuition on the definition of the covariance

correlation covariance

Charlie Parker

6,096

asked May 17, 2014 at 17:52

Stack Exchange Network

current community

your communities

more stack exchange communities

50 Bookmarks

Why square the difference instead of taking the absolute value in standard deviation?

How to normalize data to 0-1 range?

What is the trade-off between batch size and number of iterations to train a neural network?

ROC vs precision-and-recall curves

Why L1 norm for sparse models

How to compute precision/recall for multiclass-multilabel classification?

Why is the L2 regularization equivalent to Gaussian prior?

How to normalize data between -1 and 1?

What are good initial weights in a neural network?

What is the difference between ZCA whitening and PCA whitening?

What are the shortcomings of the Mean Absolute Percentage Error (MAPE)?

Clustering with K-Means and EM: how are they related?

Backpropagation with Softmax / Cross Entropy

What is the definition of a "feature map" (aka "activation map") in a convolutional neural network?

Which loss function is correct for logistic regression?

Why zero correlation does not necessarily imply independence

What are the factors that cause the posterior distributions to be intractable?

Entropy of an image

How can SVM 'find' an infinite feature space where linear separation is always possible?

What is difference between 'transfer learning' and 'domain adaptation'?

Who invented k-fold cross-validation?

Backpropagation on a convolutional layer

Backpropagation algorithm and error in hidden layer

Why do we divide by the standard deviation and not some other standardizing factor before doing PCA?

Comparing two histograms using Chi-Square distance

How to do data augmentation and train-validate split?

In CNN, are upsampling and transpose convolution the same?

Softmax overflow [closed]

What is the difference between linearly dependent and linearly correlated?

Intuition on the definition of the covariance