Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
74 views

Python Machine Learning Errata 2nd

The document contains a list of errata for various pages in machine learning textbooks and code examples. There are over 19 entries that provide corrections to technical descriptions, code snippets, equations, and other details. The errata include fixes to topics like neural network architectures, data preprocessing, model training procedures, and other algorithm explanations.

Uploaded by

Chandrasekar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
74 views

Python Machine Learning Errata 2nd

The document contains a list of errata for various pages in machine learning textbooks and code examples. There are over 19 entries that provide corrections to technical descriptions, code snippets, equations, and other details. The errata include fixes to topics like neural network architectures, data preprocessing, model training procedures, and other algorithm explanations.

Uploaded by

Chandrasekar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

Errata

- 19 submitted: last submission 08 Oct 2018

Page no. 386

In the diagram of multilayer perceptron, the hidden layer description is: "Hidden layer with 3 hidden
units plus bias unit (d = 3+1)"

It should be:

Hidden layer with 4 hidden units plus bias unit (d = 4+1)

Page no. 437

In the implementation of generating batches of data, the following line is incorrect:

yield (X[i:i+batch_size, :], y[i:i+batch_size])

It should be:

yield (X_copy[i:i+batch_size, :], y_copy[i:i+batch_size])

Page no. 468

The following paragraph is bit confusing:

"In the following example, we assume that data from source A is fed through placeholder, and
source B is the output of a generator network which we will build the generator network by calling the
build_generator function within the generator scope, then we will add a classifier by calling
build_classifier within the classifier scope:"

Consider this paragraph:

"In the following example, we assume that data from source A is fed through placeholder, and
source B is the output of a generator network. We will build the generator network by calling the
build_generator function within the generator scope, then we will add a classifier by calling
build_classifier within the classifier scope:"

Page no. 482

"t1 = tf.ones(shape=(10, 1)," should be "t1 = tf.ones(shape=5, 1),"

and
"t2 = tf.zeros(shape=(10, 1)," should be "t2 = tf.zeros(shape=(5, 1),"

Page no. 499, Chapter 15

It is:

So, you can see that this different treatment of elements of x can artifcially put more emphasize on
the middle element, x[2], since it has appeared in most computations.

It should be:

So, you can see that this different treatment of elements of x can artifcially put
more emphasis on the middle element, x[2], since it has appeared in most computations.

Page no. 517, Chapter 15

X_valid_centered = X_valid - mean_vals

This line should be:

X_valid_centered = (X_valid - mean_vals)/std_val

Page no. 518, Chapter 15

weights = tf.get_variable(name='_weights',

shape=weights_shape,

It should be:

weights = tf.get_variable(name='_weights',

shape=weights_shape)

Page no. 527, Chapter 15

The following code shows how to restore a saved mode.

It should be:

The following code shows how to restore a saved model.

Page no. 542, Chapter 16

"the activation of the same hidden layer from the previous time step t=1"

It should be:
"the activation of the same hidden layer from the previous time step t-1"

Chapter 3, Page 55

It is:

( 6 / 45 ≈ 0.067 )

It should be:

( 3 / 45 ≈ 0.067 )

Chapter 5, Page 155

It is:

Both LDA and PCA are linear transformation techniques that can be used to reduce the number of
dimensions in a dataset; the former is an unsupervised algorithm, whereas the latter is supervised.

It should be:

Both PCA and LDA are linear transformation techniques that can be used to reduce the number of
dimensions in a dataset; the former is an unsupervised algorithm, whereas the latter is supervised.

Errata Type: Typo | Page 28

It is:
sum([j * j for i, j in zip(a, b)]

Should be:
sum([i * j for i, j in zip(a, b)]

Errata type: Technical l Page no: 72

This: The first row corresponds to the class-membership probabilities of the first flower,the second
row corresponds to the class-membership probabilities of the third flower, and so forth.

Should be: The first row corresponds to the class-membership probabilities of the first flower,the
second row corresponds to the class-membership probabilities of the second flower, and so forth.

Errata Type: Code|Page Number: On page: 385

It is:

Wk,j(l)." // --> [W {subscript k,j}


{superscript(l)}].
should be:

Wk,j(l+1). //--> [ W {subscript k,j}


{superscript(l)}].

Errata Type: Code I Page number: 66

It is:

J(w) = - sum_i y^(i) * log(Phi(z^(i))) + (1 - y^(i)) * log(1 - Phi(z^(i)))

Should be:

> J(w) = - sum_i [y^(i) * log(Phi(z^(i))) + (1 - y^(i)) * log(1 - Phi(z^(i)))]

Page : 91

It is:

Here, p(i|t) is the proportion of the samples that belong to class c for a
particular node t."

should be:

Here, p(i|t) is the proportion of the samples that belong to class i for a particular node t."

Errata Type: Technical |Page Number: 385

It is:
w(l)k,j

Should be:
w(l+1)k,j

testing errata upload

Errata Type: Code | Chapter 5 | Page 159


This:
n = X_train[y_train == i + 1, :].shape[0]

Should Be:
n = X_train_std[y_train == i + 1, :].shape[0]

You might also like