Change to TenorFlow 2 from version 1

Adjust to changes in sklearn, theano, keras

Author: rrlyman

n0_network.py

@@ -1,4 +1,6 @@
-import tensorflow as tf  
+#import tensorflow as tf  
+from tensorflow.compat import v1 as tf
+tf.compat.v1.disable_eager_execution()
 import numpy as np
 from collections import namedtuple
 import datetime

n1_2cnv1fc.py

@@ -1,4 +1,6 @@
-import tensorflow as tf  
+#import tensorflow as tf  
+from tensorflow.compat import v1 as tf
+tf.compat.v1.disable_eager_execution()
 import numpy as np
 from collections import namedtuple
 import datetime
@@ -43,6 +45,7 @@ def __init__(self, truthed_features, dtype=np.float32):
                 nm = 'x_'+nm
             if i>1:
                 extra_features_width += truthed_features.feature_width[i]
+   
             lst.append(tf.placeholder(dtype, shape=[None, truthed_features.feature_width[i]], name=nm))
 
         # ph is a named tuple with key names like 'image', 'm_label', and values that
@@ -218,7 +221,7 @@ def computeSize(s,tens):
             tShape = tens.get_shape()
             nDims = len(tShape)
             for i in range(nDims):
-                sumC *= tShape[i].value
+                sumC *= tShape[i]
             print ('\t{}\t{}'.format(s,sumC),flush=True)
             return sumC
 

n1_2cnv2fc.py

@@ -1,4 +1,6 @@
-import tensorflow as tf  
+#import tensorflow as tf  
+from tensorflow.compat import v1 as tf
+tf.compat.v1.disable_eager_execution()
 import numpy as np
 from collections import namedtuple
 import datetime
@@ -250,7 +252,7 @@ def computeSize(s,tens):
             tShape = tens.get_shape()
             nDims = len(tShape)
             for i in range(nDims):
-                sumC *= tShape[i].value
+                sumC *= tShape[i]
             print ('\t{}\t{}'.format(s,sumC),flush=True)
             return sumC
 

n1_baseTensorNN.py

@@ -1,4 +1,6 @@
-import tensorflow as tf  
+from tensorflow.compat import v1 as tf
+tf.compat.v1.disable_eager_execution()
+#import tf
 import numpy as np
 from collections import namedtuple
 import datetime

n1_image_to_image.py

@@ -1,4 +1,5 @@
-import tensorflow as tf  
+from tensorflow.compat import v1 as tf
+#import tf
 import numpy as np
 from collections import namedtuple
 import datetime
@@ -258,7 +259,7 @@ def computeSize(s,tens):
             tShape = tens.get_shape()
             nDims = len(tShape)
             for i in range(nDims):
-                sumC *= tShape[i].value
+                sumC *= tShape[i]
             print ('\t{}\t{}'.format(s,sumC),flush=True)
             return sumC
 

n1_residual3x4.py

@@ -7,7 +7,7 @@
 
 
 '''
-import tensorflow as tf  
+from tensorflow.compat import v1 as tf
 import numpy as np
 from collections import namedtuple
 import datetime
@@ -91,7 +91,7 @@ def bias_variable(shape, dtype):
             return tf.Variable(initial)   
 
         def shapeOuts(n):
-            print ('n={}, hin={},w={}, b={} ,hout={}\n'.format(n, h[n]._shape, w[n]._variable._shape, b[n]._variable._shape, h[n+1]._shape))
+            print ('n={}, hin={},w={}, b={} ,hout={}\n'.format(n, h[n].shape, w[n].shape, b[n].shape, h[n+1]._shape))
 
         def section(n):
             with tf.name_scope('section_'+str(n)+'_0') as scope:     
@@ -122,7 +122,7 @@ def computeSize(s,tens):
             tShape = tens.get_shape()
             nDims = len(tShape)
             for i in range(nDims):
-                sumC *= tShape[i].value
+                sumC *= tShape[i]
             print ('\t{}\t{}'.format(s,sumC),flush=True)
             return sumC
 

o1_top_secret_cnn.py

@@ -44,6 +44,10 @@
 #import n1_2cnv1fc as nnetwork     
 #import n1_residual3x4 as nnetwork 
 import n1_2cnv2fc as nnetwork     
+import skimage.transform as af  
+from bitarray import bitarray 
+
+
 input_filters_dict = {'m_label': list(range(48,58))+list(range(65,91))}  
 output_feature_list  = ['orientation_one_hot','image']   
 dtype = np.float32
@@ -70,7 +74,7 @@
 
 
 # pick up the base characters from training_image_file
-# produce some skeared versions 
+# produce some sheared versions 
 # make into a training set
 # place in a ocr_utils TruthedCharacters class so we can use the
 # one hot and batch functions 
@@ -94,7 +98,7 @@
 orientation=[]
 recognized_label =[]
 
-import skimage.transform as af  
+
 
 for j in range(shp[0]):
     for i,skew in enumerate(skewRange):
@@ -141,10 +145,10 @@
 image_file_jpg = image_file+'.jpg'
 df,t1 = ocr_utils.file_to_df(image_file,character_size, title = 'unencrypted file',white_space=white_space)
 
-from bitarray import bitarray  
+ 
 secret_message = "top secret"
 a = bitarray()   
-a.fromstring(secret_message)
+a.frombytes(secret_message.encode('utf_8'))
 
 index = 0
 encoded_skews=[]

o2_top_secret_lda-tesseract.py

@@ -12,6 +12,7 @@
 from sklearn.linear_model import LogisticRegression
 from sklearn.discriminant_analysis import LinearDiscriminantAnalysis as LDA   
 from sklearn.metrics import accuracy_score
+from ruamel_yaml.compat import utf8
 
 inputs = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghiklnopqrstuvwxyz'
 inputs_list = list(ord(x) for x in inputs)
@@ -70,8 +71,7 @@ def encode_and_save_file(input_base, output_base, character_size, white_space, s
 
     from bitarray import bitarray  
     a = bitarray()   
-    a.fromstring(secret_message)
-    
+    a.frombytes(secret_message.encode('utf-8') )   
     index = 0
 
     def convert_to_shear(a):

o3_top_secret_python_box.py

@@ -73,7 +73,7 @@
 #from sklearn.model_selection
 from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
 from sklearn.linear_model import LogisticRegression
-from sklearn.cross_validation import train_test_split
+from sklearn.model_selection import train_test_split
 
 # input_filters_dict = {'m_label': list(range(48,58))+list(range(65,91))}  
 # output_feature_list  = ['orientation_one_hot','image']   

o4_image_to_image.py

@@ -30,7 +30,7 @@
 import pandas as pd
 import n1_image_to_image as nnetwork  
 #import n1_residual3x4 as nnetwork  
-import tensorflow as tf  
+from tensorflow.compat import v1 as tf 
 dtype = np.float32
 #with tf.device('/GPU:0'):
 #with tf.device('/cpu:0'): 

ocr_utils.py

@@ -38,7 +38,8 @@
 import numpy as np
 import pandas as pd
 import math
-from pandas.io.common import ZipFile
+#from pandas.io.common import ZipFile
+from zipfile import ZipFile
 from matplotlib.colors import ListedColormap
 import matplotlib.pyplot as plt
 import sys
@@ -1064,9 +1065,13 @@ def plot_decision_regions(X=None, y=None, classifier=None, resolution = .005, te
     plt.ylim(xx2.min()-d, xx2.max()+d)
 
     # plot class samples
+
     for idx, cl in enumerate(np.unique(y)):
-        plt.scatter(X[y == cl, 0], X[y == cl, 1],
-                    alpha=0.8, c=cmap(idx),
+        xs = X[y == cl, 0]
+        ys = X[y == cl, 1]
+        c =cmap(idx)
+        plt.scatter(xs, ys,
+                    alpha=0.8, color=c,
                     marker=markers[idx%len(markers)], label=cl) 
 
     # highlight test samples

p115_l1_l2_regularization.py

@@ -69,7 +69,7 @@
 
 from sklearn.linear_model import LogisticRegression
 
-lr = LogisticRegression(penalty='l1', C=0.1, random_state=0)
+lr = LogisticRegression(penalty='l1', C=0.1, random_state=0, solver='liblinear',multi_class='auto')
 lr.fit(X_train_std, y_train)
 print('Training accuracy-l1 regularization:', lr.score(X_train_std, y_train))
 print('Test accuracy-l1 regularization:', lr.score(X_test_std, y_test))
@@ -79,7 +79,7 @@
 print('\t{}'.format(lr.coef_))
 
 
-lr = LogisticRegression(penalty='l2', C=0.1, random_state=0)
+lr = LogisticRegression(penalty='l2', C=0.1, random_state=0, solver='liblinear',multi_class='auto')
 lr.fit(X_train_std, y_train)
 print('Training accuracy-l2 regularization:', lr.score(X_train_std, y_train))
 print('Test accuracy-l2 regularization:', lr.score(X_test_std, y_test))
@@ -101,7 +101,7 @@
 def weight_graph(regularization = 'l1'):
     weights, params = [], []
     for c in np.arange(0, 6):
-        lr = LogisticRegression(penalty=regularization, C=10**c, random_state=0)
+        lr = LogisticRegression(penalty=regularization, C=10**c, random_state=0, solver='liblinear',multi_class='auto')
         lr.fit(X_train_std, y_train)
         weights.append(lr.coef_[1])
         params.append(10**c)

p119_squential_backward_selection.py

@@ -49,7 +49,7 @@
 y, X, y_test,  X_test, labels  = ocr_utils.load_E13B(chars_to_train = (48,49,50) , columns=range(0,20), nChars=1000, random_state=0) 
 
 
-from sklearn.cross_validation import train_test_split
+from sklearn.model_selection import train_test_split
 
 X_train, X_test, y_train, y_test = train_test_split(
          X, y, test_size=0.3, random_state=0)

p131_principal_component_analysis.py

@@ -130,7 +130,7 @@
 X_train_pca = pca.fit_transform(X_train_image)
 X_test_pca = pca.transform(X_test_image)
 
-lr = LogisticRegression()
+lr = LogisticRegression(solver='liblinear',multi_class='auto')
 logistic_fitted =lr.fit(X_train_pca, y_train)
 
 print('\nPCA Train Accuracy: {:4.6f}, n_components={}'.format(accuracy_score(y_train, logistic_fitted.predict(X_train_pca)),pca.n_components))
@@ -149,7 +149,7 @@
 X_train_pca = pca.fit_transform(X_train_image)
 X_test_pca = pca.transform(X_test_image)
 
-lr = LogisticRegression()
+lr = LogisticRegression(solver='liblinear',multi_class='auto')
 logistic_fitted = lr.fit(X_train_pca, y_train)
 
 y_train_pred = logistic_fitted.predict(X_train_pca)
@@ -191,7 +191,7 @@
 X_train_pca = pca.fit_transform(X_train_image)
 X_test_pca = pca.transform(X_test_image)
 
-lr = LogisticRegression()
+lr = LogisticRegression(solver='liblinear',multi_class='auto')
 logistic_fitted=lr.fit(X_train_pca, y_train)
 y_train_pred = logistic_fitted.predict(X_train_pca)
 y_test_pred = logistic_fitted.predict(X_test_pca)

p141_linear_descriminant_analsys.py

@@ -61,6 +61,7 @@
 from sklearn.discriminant_analysis import  LinearDiscriminantAnalysis as LDA   
 print_limit = 20
 chars_to_train = range(48,58)
+n_classes = len(chars_to_train)
 columnsXY=range(0,20)
 column_str = 'column_sum{}'.format(list(columnsXY))
 
@@ -107,7 +108,7 @@
 S_W = np.zeros((d, d))
 for label, mv in zip(unique_labels, mean_vecs):
     class_scatter = np.zeros((d, d))
-    for row in X_train_std[[y_train == label]]:
+    for row in X_train_std[y_train == label]:
         row, mv = row.reshape(d, 1), mv.reshape(d, 1)
         class_scatter += (row-mv).dot((row-mv).T)
     S_W += class_scatter
@@ -195,7 +196,7 @@
 X_test_lda =  lda.transform(X_test_std)
 
 from sklearn.linear_model import LogisticRegression
-lr = LogisticRegression()
+lr = LogisticRegression(solver='liblinear', multi_class='auto')
 lr = lr.fit(X_train_lda, y_train)
 
 title = 'Linear Descriminant Analysis Training Set'
@@ -208,13 +209,13 @@
 
 ###############################################################################
 n_components = 10
-lda = LDA(n_components=n_components)
+lda = LDA(n_components=min(n_components,n_classes-1))
 X_train_lda = lda.fit_transform(X_train_std, y_train)
 X_test_lda = lda.transform(X_test_std)
 
 print ('n_components={}'.format(lda.n_components))
 
-lr = LogisticRegression()
+lr = LogisticRegression(solver='liblinear', multi_class='auto')
 logistic_fitted = lr.fit(X_train_lda, y_train)
 
 from sklearn.metrics import accuracy_score
@@ -233,13 +234,13 @@
 
 ###############################################################################
 n_components = 10
-lda = LDA(n_components=n_components, solver='eigen')
+lda = LDA(n_components=n_components-1, solver='eigen')
 X_train_lda = lda.fit_transform(X_train_std, y_train)
 X_test_lda = lda.transform(X_test_std)
 
 print ('n_components={}'.format(lda.n_components))
 
-lr = LogisticRegression()
+lr = LogisticRegression(solver='liblinear', multi_class='auto')
 logistic_fitted = lr.fit(X_train_lda, y_train)
 
 from sklearn.metrics import accuracy_score

p154_pca_nonlinear_mapings.py

@@ -83,8 +83,7 @@ def rbf_kernel_pca1(X, gamma, n_components):
     eigvals, eigvecs = eigh(K)
 
     # Collect the top k eigenvectors (projected samples)
-    X_pc = np.column_stack((eigvecs[:, -i]
-                            for i in range(1, n_components + 1)))
+    X_pc = np.column_stack([eigvecs[:, -i] for i in range(1, n_components + 1)])
 
     return X_pc
 
@@ -269,7 +268,7 @@ def rbf_kernel_pca(X, gamma, n_components):
     eigvals, eigvecs = eigh(K)
 
     # Collect the top k eigenvectors (projected samples)
-    alphas = np.column_stack((eigvecs[:,-i] for i in range(1,n_components+1)))
+    alphas = np.column_stack([eigvecs[:,-i] for i in range(1,n_components+1)])
 
     # Collect the corresponding eigenvalues
     lambdas = [eigvals[-i] for i in range(1,n_components+1)]