4/22/24, 1:23 PM DL 2 .

ipynb - Colab

# Name: Malhar Badame

# Roll No.: COBA010
# Lab Assignment : 2

# Importing Necesarry Packages

import numpy as np
from numpy.ma.core import argmax
import pandas as pd
from matplotlib import cm
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns
#import os
import time
from sklearn.metrics import confusion_matrix, accuracy_score, auc
from keras.preprocessing import sequence
from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation
from keras.layers import Embedding
from keras.layers import Conv1D, GlobalMaxPooling1D
from keras.callbacks import EarlyStopping
from keras import models
from keras import layers
from keras.datasets import imdb

# Loading the dataset

(X_train, y_train), (X_test, y_test) = imdb.load_data()
X = np.concatenate((X_train, X_test), axis=0)
y = np.concatenate((y_train, y_test), axis=0)

# Exploring the Data

print("Training data: ")
print(X.shape)
print(y.shape)
print("Classes: ")
print(np.unique(y))
print("Number of words: ")
print(len(np.unique(np.hstack(X))))
print("Review length: ")
result = [len(x) for x in X]
print("Mean %.2f words (%f)" % (np.mean(result), np.std(result))) # Ploting the review length
plt.boxplot(result)
plt.show()

Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/imd

17464789/17464789 [==============================] - 0s 0us/step
Training data:
(50000,)
(50000,)
Classes:
[0 1]
Number of words:
88585
Review length:
Mean 234.76 words (172.911495)

https://colab.research.google.com/drive/10xWLhOwYzGpiVz27yd5yOAxDpXpFaGnN#printMode=true 1/6
4/22/24, 1:23 PM DL 2 .ipynb - Colab
def vectorize_sequences(sequences, dimension=5000): # Function for vectorising data
results = np.zeros((len(sequences), dimension)) # Creating an all-zero matrix of shape (len(sequences), dimension)
for i, sequence in enumerate(sequences):
results[i, sequence] = 1. # Set specific indices of results[i] to 1s
return results

# Creating Training and Testing Sets and Preprocessing them

(train_data, train_labels), (test_data, test_labels) = imdb.load_data(num_words=5000)
# Our vectorized training data
x_train = vectorize_sequences(train_data)
# Our vectorized test data
x_test = vectorize_sequences(test_data)
# Our vectorized labels one-hot encoder
y_train = np.asarray(train_labels).astype('float32')
y_test = np.asarray(test_labels).astype('float32')

# Creating the DNN Model

model = models.Sequential()
model.add(layers.Dense(32, activation='relu', input_shape=(5000,)))
model.add(layers.Dense(32, activation='relu',))
model.add(layers.Dense(1, activation='sigmoid'))

#Set validation set aside

x_val = x_train[:10000]
partial_x_train = x_train[10000:]
y_val = y_train[:10000]
partial_y_train = y_train[10000:]

# Compiling Model
model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['acc'])
start_time_m1 = time.time()
history = model.fit(partial_x_train,
partial_y_train,
epochs=20,
batch_size=512,
validation_data=(x_val, y_val))
total_time_m1 = time.time() - start_time_m1
print("The Dense Convolutional Neural Network 1 layer took %.4f seconds to train." % (total_time_m1))

Epoch 1/20
30/30 [==============================] - 3s 79ms/step - loss: 0.5512 - acc: 0.7727 - val_loss: 0.3889 - val_acc: 0.8479
Epoch 2/20
30/30 [==============================] - 1s 45ms/step - loss: 0.2926 - acc: 0.8936 - val_loss: 0.2950 - val_acc: 0.8816
Epoch 3/20
30/30 [==============================] - 1s 31ms/step - loss: 0.2108 - acc: 0.9213 - val_loss: 0.2880 - val_acc: 0.8831
Epoch 4/20
30/30 [==============================] - 1s 32ms/step - loss: 0.1706 - acc: 0.9383 - val_loss: 0.3042 - val_acc: 0.8783
Epoch 5/20
30/30 [==============================] - 1s 31ms/step - loss: 0.1420 - acc: 0.9514 - val_loss: 0.3253 - val_acc: 0.8748
Epoch 6/20
30/30 [==============================] - 1s 34ms/step - loss: 0.1208 - acc: 0.9587 - val_loss: 0.3621 - val_acc: 0.8721
Epoch 7/20
30/30 [==============================] - 1s 35ms/step - loss: 0.1010 - acc: 0.9669 - val_loss: 0.3928 - val_acc: 0.8659
Epoch 8/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0830 - acc: 0.9743 - val_loss: 0.4288 - val_acc: 0.8636
Epoch 9/20
30/30 [==============================] - 1s 32ms/step - loss: 0.0655 - acc: 0.9817 - val_loss: 0.4705 - val_acc: 0.8608
Epoch 10/20
30/30 [==============================] - 1s 30ms/step - loss: 0.0520 - acc: 0.9885 - val_loss: 0.5118 - val_acc: 0.8567
Epoch 11/20
30/30 [==============================] - 1s 31ms/step - loss: 0.0395 - acc: 0.9923 - val_loss: 0.5597 - val_acc: 0.8559
Epoch 12/20
30/30 [==============================] - 1s 31ms/step - loss: 0.0283 - acc: 0.9965 - val_loss: 0.6085 - val_acc: 0.8540
Epoch 13/20
30/30 [==============================] - 2s 52ms/step - loss: 0.0207 - acc: 0.9981 - val_loss: 0.6437 - val_acc: 0.8552
Epoch 14/20
30/30 [==============================] - 2s 56ms/step - loss: 0.0139 - acc: 0.9996 - val_loss: 0.6840 - val_acc: 0.8537
Epoch 15/20
30/30 [==============================] - 1s 31ms/step - loss: 0.0095 - acc: 0.9997 - val_loss: 0.7194 - val_acc: 0.8526
Epoch 16/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0067 - acc: 0.9999 - val_loss: 0.7480 - val_acc: 0.8541
Epoch 17/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0049 - acc: 0.9999 - val_loss: 0.7748 - val_acc: 0.8521
Epoch 18/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0038 - acc: 1.0000 - val_loss: 0.7988 - val_acc: 0.8520
Epoch 19/20
30/30 [==============================] - 1s 31ms/step - loss: 0.0031 - acc: 1.0000 - val_loss: 0.8184 - val_acc: 0.8534
Epoch 20/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0025 - acc: 1.0000 - val_loss: 0.8367 - val_acc: 0.8526
The Dense Convolutional Neural Network 1 layer took 24.8242 seconds to train.

history_dict = history.history
history_dict.keys()

https://colab.research.google.com/drive/10xWLhOwYzGpiVz27yd5yOAxDpXpFaGnN#printMode=true 2/6
4/22/24, 1:23 PM DL 2 .ipynb - Colab
y_ y ()

dict_keys(['loss', 'acc', 'val_loss', 'val_acc'])

acc = history.history['acc']
val_acc = history.history['val_acc']
loss = history.history['loss']
val_loss = history.history['val_loss']
epochs = range(1, len(acc) + 1)

# Plotting model loss

plt.plot(epochs, loss, 'bo', label='Training loss') # "bo" is for "blue dot"
plt.plot(epochs, val_loss, 'b', label='Validation loss') # b is for "solid blue line"
plt.title('Training and validation loss')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.show()

plt.clf() # clear figure

acc_values = history_dict['acc']
val_acc_values = history_dict['val_acc']

# Plotting model accuracy

plt.plot(epochs, acc, 'bo', label='Training acc')
plt.plot(epochs, val_acc, 'b', label='Validation acc')
plt.title('Training and validation accuracy')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.show()

https://colab.research.google.com/drive/10xWLhOwYzGpiVz27yd5yOAxDpXpFaGnN#printMode=true 3/6
4/22/24, 1:23 PM DL 2 .ipynb - Colab
# Model Summary
print(model.summary())

# Predictions
pred = model.predict(x_test)
classes_x=np.argmax(pred,axis=1)
accuracy_score(y_test,classes_x)

#Confusion Matrix
conf_mat = confusion_matrix(y_test, classes_x)
print(conf_mat)

conf_mat_normalized = conf_mat.astype('float') / conf_mat.sum(axis=1)[:, np.newaxis]

sns.heatmap(conf_mat_normalized)
plt.ylabel('True label')
plt.xlabel('Predicted label')

Model: "sequential"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense (Dense) (None, 32) 160032

dense_1 (Dense) (None, 32) 1056

dense_2 (Dense) (None, 1) 33

=================================================================
Total params: 161121 (629.38 KB)
Trainable params: 161121 (629.38 KB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________
None
782/782 [==============================] - 3s 3ms/step
[[12500 0]
[12500 0]]
Text(0.5, 23.52222222222222, 'Predicted label')

#Dense with Two Layer

model2 = models.Sequential()
model2.add(layers.Dense(32, activation='relu', input_shape=(5000,)))
model2.add(layers.Dense(32, activation='relu'))
model2.add(layers.Dense(32, activation='relu'))
model2.add(layers.Dense(1, activation='sigmoid'))

# Compiling Model
model2.compile(optimizer='adam', loss='binary_crossentropy', metrics=['acc'])
start_time_m2 = time.time()
history= model2.fit(partial_x_train,
partial_y_train,
epochs=20,
batch_size=512,
validation_data=(x_val, y_val))
total_time_m2 = time.time() - start_time_m2
print("The Dense Convolutional Neural Network 2 layers took %.4f seconds to train." % (total_time_m2))

Epoch 1/20
30/30 [==============================] - 3s 61ms/step - loss: 0.5641 - acc: 0.7379 - val_loss: 0.3708 - val_acc: 0.8590
Epoch 2/20
30/30 [==============================] - 1s 34ms/step - loss: 0.2742 - acc: 0.8943 - val_loss: 0.2890 - val_acc: 0.8842

https://colab.research.google.com/drive/10xWLhOwYzGpiVz27yd5yOAxDpXpFaGnN#printMode=true 4/6
4/22/24, 1:23 PM DL 2 .ipynb - Colab
Epoch 3/20
30/30 [==============================] - 1s 31ms/step - loss: 0.1906 - acc: 0.9304 - val_loss: 0.3050 - val_acc: 0.8804
Epoch 4/20
30/30 [==============================] - 1s 30ms/step - loss: 0.1464 - acc: 0.9475 - val_loss: 0.3439 - val_acc: 0.8716
Epoch 5/20
30/30 [==============================] - 1s 31ms/step - loss: 0.1126 - acc: 0.9623 - val_loss: 0.3771 - val_acc: 0.8685
Epoch 6/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0817 - acc: 0.9743 - val_loss: 0.4260 - val_acc: 0.8664
Epoch 7/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0506 - acc: 0.9870 - val_loss: 0.4980 - val_acc: 0.8636
Epoch 8/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0286 - acc: 0.9951 - val_loss: 0.5630 - val_acc: 0.8620
Epoch 9/20
30/30 [==============================] - 2s 57ms/step - loss: 0.0141 - acc: 0.9989 - val_loss: 0.6416 - val_acc: 0.8596
Epoch 10/20
30/30 [==============================] - 1s 42ms/step - loss: 0.0074 - acc: 0.9997 - val_loss: 0.6919 - val_acc: 0.8603
Epoch 11/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0040 - acc: 0.9999 - val_loss: 0.7355 - val_acc: 0.8590
Epoch 12/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0025 - acc: 1.0000 - val_loss: 0.7728 - val_acc: 0.8603
Epoch 13/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0017 - acc: 1.0000 - val_loss: 0.8039 - val_acc: 0.8594
Epoch 14/20
30/30 [==============================] - 1s 33ms/step - loss: 0.0013 - acc: 1.0000 - val_loss: 0.8279 - val_acc: 0.8599
Epoch 15/20
30/30 [==============================] - 1s 31ms/step - loss: 0.0010 - acc: 1.0000 - val_loss: 0.8495 - val_acc: 0.8607
Epoch 16/20
30/30 [==============================] - 1s 33ms/step - loss: 8.3106e-04 - acc: 1.0000 - val_loss: 0.8688 - val_acc: 0.8599
Epoch 17/20
30/30 [==============================] - 1s 34ms/step - loss: 6.9839e-04 - acc: 1.0000 - val_loss: 0.8877 - val_acc: 0.8594
Epoch 18/20
30/30 [==============================] - 1s 33ms/step - loss: 5.8774e-04 - acc: 1.0000 - val_loss: 0.9059 - val_acc: 0.8594
Epoch 19/20
30/30 [==============================] - 1s 33ms/step - loss: 5.0707e-04 - acc: 1.0000 - val_loss: 0.9190 - val_acc: 0.8598
Epoch 20/20
30/30 [==============================] - 1s 29ms/step - loss: 4.4136e-04 - acc: 1.0000 - val_loss: 0.9329 - val_acc: 0.8597
The Dense Convolutional Neural Network 2 layers took 23.3940 seconds to train.

acc = history.history['acc']
val_acc = history.history['val_acc']
loss = history.history['loss']
val_loss = history.history['val_loss']
epochs = range(1, len(acc) + 1)

# Plotting Loss
plt.plot(epochs, loss, 'bo', label='Training loss') # "bo" is for "blue dot"
plt.plot(epochs, val_loss, 'b', label='Validation loss') # b is for "solid blue line"
plt.title('DNN 2 layer Training and validation loss')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.show()

https://colab.research.google.com/drive/10xWLhOwYzGpiVz27yd5yOAxDpXpFaGnN#printMode=true 5/6
4/22/24, 1:23 PM DL 2 .ipynb - Colab
plt.clf() # clear figure
acc_values = history_dict['acc']
val_acc_values = history_dict['val_acc']
# Plotting Accuracy
plt.plot(epochs, acc, 'bo', label='Training acc')
plt.plot(epochs, val_acc, 'b', label='Validation acc')
plt.title('DNN 2 layer Training and validation accuracy')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.show()

print(model2.summary())
# Predictions
pred = model2.predict(x_test)
classes_x=np.argmax(pred,axis=-1)
accuracy_score(y_test,classes_x)

output Model: "sequential_1"

_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
dense_3 (Dense) (None, 32) 160032

dense_4 (Dense) (None, 32) 1056

dense_5 (Dense) (None, 32) 1056

dense_6 (Dense) (None, 1) 33

=================================================================
Total params: 162177 (633.50 KB)
Trainable params: 162177 (633.50 KB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________
None
782/782 [==============================] - 2s 2ms/step
0.5

https://colab.research.google.com/drive/10xWLhOwYzGpiVz27yd5yOAxDpXpFaGnN#printMode=true 6/6