Added the final code file

main_code.py

@@ -0,0 +1,162 @@
+# -*- coding: utf-8 -*-
+"""IE_PROJECT.ipynb
+
+Automatically generated by Colaboratory.
+
+Original file is located at
+    https://colab.research.google.com/drive/1R8YipvIIF4_mZOYzJSR6UTEj7xHCiUtz
+"""
+
+from keras.datasets import mnist
+import matplotlib.pyplot as plt
+
+(X_train, y_train), (X_test, y_test) = mnist.load_data()
+X_set=X_test
+import numpy as np
+
+from keras.models import Sequential
+from keras.layers import Convolution2D as Conv2D
+from keras.layers import MaxPooling2D,Activation,Dropout
+from keras.layers import Flatten
+from keras.layers import Dense
+from keras.utils import np_utils
+from keras.layers.normalization import BatchNormalization
+
+
+X_train=X_train.reshape(-1,28,28,1)/255.0
+X_test=X_test.reshape(-1,28,28,1)/255.0
+
+y_train = np_utils.to_categorical(y_train)
+y_test = np_utils.to_categorical(y_test)
+
+import torch
+from torchvision import models
+# Load alexnet model
+alexnet = models.alexnet(pretrained=True)
+print(alexnet)
+
+# My implementation of AlexNet using keras
+model = Sequential()
+
+# (0)
+model.add(Conv2D(filters=96, input_shape=(28,28,1), kernel_size=(3,3),strides=(4,4), padding='valid'))
+
+# (1)
+model.add(Activation('relu'))
+
+# (2)
+model.add(MaxPooling2D(pool_size=(1,1), strides=(2,2), padding='valid'))
+
+# Normalising 
+model.add(BatchNormalization())
+
+# (3)
+model.add(Conv2D(filters=256, kernel_size=(3,3), strides=(1,1), padding='valid'))
+
+# (4)
+model.add(Activation('relu'))
+
+
+# (5)
+model.add(MaxPooling2D(pool_size=(1,1), strides=(2,2), padding='valid'))
+
+# Normalising
+model.add(BatchNormalization())
+
+# (6)
+model.add(Conv2D(filters=384, kernel_size=(1,1), strides=(1,1), padding='valid'))
+
+# (7)
+model.add(Activation('relu'))
+
+
+# Normalisng
+model.add(BatchNormalization())
+
+# (8)
+model.add(Conv2D(filters=384, kernel_size=(1,1), strides=(1,1), padding='valid'))
+
+#(9)
+model.add(Activation('relu'))
+
+
+#Normalising
+model.add(BatchNormalization())
+
+# (10)
+model.add(Conv2D(filters=256, kernel_size=(1,1), strides=(1,1), padding='valid'))
+
+#(11)
+model.add(Activation('relu'))
+
+# (12)
+model.add(MaxPooling2D(pool_size=(1,1), strides=(2,2), padding='valid'))
+
+
+# Normalising
+model.add(BatchNormalization())
+
+
+# AdaptiveAvgPool2d
+model.add(Flatten())
+model.add(Dense(4096, input_shape=(28*28*1,)))
+
+model.add(Activation('relu'))
+
+# (0)
+model.add(Dropout(0.5))
+
+#Normalising
+model.add(BatchNormalization())
+# (1)
+model.add(Dense(4096))
+
+# (2)
+model.add(Activation('relu'))
+
+# (3)
+model.add(Dropout(0.5))
+
+# Normalising
+model.add(BatchNormalization())
+
+# (4)
+model.add(Dense(1000))
+# (5)
+model.add(Activation('relu'))
+
+
+
+
+# Normalising
+model.add(BatchNormalization())
+
+# (6)
+model.add(Dense(10))
+model.add(Activation('softmax'))
+
+# AlexNet model ends here
+
+model.summary()
+
+# Compile step
+model.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])
+
+
+# Fitting step
+model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=3, batch_size=200, verbose=2)
+
+
+"""
+  Output
+
+Train on 60000 samples, validate on 10000 samples
+Epoch 1/3
+ - 355s - loss: 0.8524 - acc: 0.7469 - val_loss: 0.6196 - val_acc: 0.8106
+Epoch 2/3
+ - 350s - loss: 0.5407 - acc: 0.8253 - val_loss: 0.5186 - val_acc: 0.8397
+Epoch 3/3
+ - 351s - loss: 0.4828 - acc: 0.8433 - val_loss: 0.4859 - val_acc: 0.8416
+
+ I've included the output here because it takes around 16-17 minutes to run
+ """