Scribd
Upload
2 views

assignment2

The document outlines a process for classifying images from the Fashion MNIST dataset using both Fully Connected Neural Networks (FCNN) and Convolutional Neural Networks (CNN). It includes steps for loading and preprocessing the data, defining the neural network architectures, compiling the models, and evaluating their performance. The final results show test accuracies for both models after training.

Uploaded by

Rizwan Sanny
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
2 views

assignment2

The document outlines a process for classifying images from the Fashion MNIST dataset using both Fully Connected Neural Networks (FCNN) and Convolutional Neural Networks (CNN). It includes steps for loading and preprocessing the data, defining the neural network architectures, compiling the models, and evaluating their performance. The final results show test accuracies for both models after training.

Uploaded by

Rizwan Sanny
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Problem 2: Fashion MNIST Classification

Link to Colab Notebook: colabl ink

Import Necessary Modules


from tensorflow . keras . datasets import fashion_mnist
import matplotlib . pyplot as plt
import numpy as np
from tensorflow . keras . utils import to_categ orical
from tensorflow . keras . layers import Input , Flatten , Dense , Conv2D ,
MaxPooling2D
from tensorflow . keras . models import Model
from tensorflow . keras . optimizers import RMSprop

Load and Investigate Data


( trainX , trainY ) , ( testX , testY ) = fashion_mnist . load_data ()

print ( f ’ trainX . shape : { trainX . shape } , trainY . shape : { trainY . shape } ,


testX . shape : { testX . shape } , testY . shape : { testY . shape } ’)
print ( f ’ trainX . dtype : { trainX . dtype } , trainY . dtype : { trainY . dtype } ,
testX . dtype : { testX . dtype } , testY . dtype : { testY . dtype } ’)
print ( f ’ trainX . Range : { trainX . max () } - { trainX . min () } , testX . Range :
{ testX . max () } - { testX . min () } ’)

Display Some Images


def display_img ( img_set , title_set ) :
n = len ( title_set )
for i in range ( n ) :
plt . subplot (3 , 3 , i + 1)
plt . imshow ( img_set [ i ] , cmap = ’ gray ’)
plt . title ( title_set [ i ])
plt . show ()

display_img ( trainX [:9] , trainY [:9])

1
Prepare Dataset
# Reshape the data to add a channel dimension for CNN
trainX = np . expand_dims ( trainX , axis = -1)
testX = np . expand_dims ( testX , axis = -1)

# Normalize the data to the range [0 , 1]


trainX = trainX / 255.0
testX = testX / 255.0

# Convert labels to one - hot encoding


trainY = to_cate gorical ( trainY , num_classes =10)
testY = to_ca tegoric al ( testY , num_classes =10)

# Investigate the updated shapes


print ( f ’ trainX . shape : { trainX . shape } , trainY . shape : { trainY . shape } ,
testX . shape : { testX . shape } , testY . shape : { testY . shape } ’)

Fully Connected Neural Network (FCNN)


numClasses = 10
inputs = Input ((28 , 28 , 1) , name = " InputLayer " )

x = Flatten () ( inputs )
x = Dense (64 , activation = ’ sigmoid ’) ( x )
x = Dense (128 , activation = ’ sigmoid ’) ( x )
x = Dense (128 , activation = ’ sigmoid ’) ( x )
x = Dense (64 , activation = ’ sigmoid ’) ( x )

outputs = Dense ( numClasses , activation = ’ softmax ’ , name = ’ OutputLayer


’) ( x )

model = Model ( inputs , outputs , name = ’ FCNN ’)


model . summary ()

model . compile ( loss = ’ c a t e g o r i c a l _ c r o s s e n t r o p y ’ , optimizer = RMSprop () ,


metrics =[ ’ accuracy ’ ])
model . fit ( trainX , trainY , epochs =10 , batch_size =128)

test_loss , fcnn_test_acc = model . evaluate ( testX , testY )


print ( f ’ Test Accuracy : { fcnn_test_acc } ’)

Convolutional Neural Network (CNN)


numClasses = 10
inputs = Input ((28 , 28 , 1) )

x = Conv2D (32 , (3 , 3) , activation = ’ relu ’ , padding = ’ same ’) ( inputs )


x = MaxPooling2D ((2 , 2) ) ( x )
x = Conv2D (64 , (3 , 3) , activation = ’ relu ’ , padding = ’ same ’) ( x )

2
x = MaxPooling2D ((2 , 2) ) ( x )
x = Conv2D (128 , (3 , 3) , activation = ’ relu ’ , padding = ’ same ’) ( x )
x = MaxPooling2D ((2 , 2) ) ( x )
x = Flatten () ( x )

x = Dense (64 , activation = ’ sigmoid ’) ( x )


x = Dense (128 , activation = ’ sigmoid ’) ( x )
x = Dense (128 , activation = ’ sigmoid ’) ( x )
x = Dense (64 , activation = ’ sigmoid ’) ( x )

outputs = Dense ( numClasses , activation = ’ softmax ’ , name = ’ OutputLayer


’) ( x )

model = Model ( inputs , outputs , name = ’ CNN ’)


model . summary ()

model . compile ( loss = ’ c a t e g o r i c a l _ c r o s s e n t r o p y ’ , optimizer = RMSprop () ,


metrics =[ ’ accuracy ’ ])
model . fit ( trainX , trainY , epochs =10 , batch_size =32 ,
v a l i d a t i o n _ s p li t =0.1)

test_loss , cnn_test_acc = model . evaluate ( testX , testY )


print ( f ’ Test Accuracy : { cnn_test_acc } ’)

You might also like