Standard (Fully Connected) Neural Network¶
In [1]:
#Use in Markup cell type
#
Implementing Fully connected Neural Net¶
Loading Required packages and Data¶
In [2]:
###1. Load Data and Splot Data
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers.core import Dense, Activation
from keras.utils import np_utils
(X_train, Y_train), (X_test, Y_test) = mnist.load_data()
Using TensorFlow backend.
Preprocessing¶
In [3]:
import matplotlib.pyplot as plt
n = 10 # how many digits we will display
plt.figure(figsize=(20, 4))
for i in range(n):
# display original
ax = plt.subplot(2, n, i + 1)
plt.imshow(X_test[i].reshape(28, 28))
plt.gray()
ax.get_xaxis().set_visible(False)
ax.get_yaxis().set_visible(False)
plt.show()
plt.close()
<Figure size 2000x400 with 10 Axes>
In [4]:
print("Previous X_train shape: {} \nPrevious Y_train shape:{}".format(X_train.shape, Y_train.shape))
X_train = X_train.reshape(60000, 784)
X_test = X_test.reshape(10000, 784)
X_train = X_train.astype('float32')
X_test = X_test.astype('float32')
X_train /= 255
X_test /= 255
classes = 10
Y_train = np_utils.to_categorical(Y_train, classes)
Y_test = np_utils.to_categorical(Y_test, classes)
print("New X_train shape: {} \nNew Y_train shape:{}".format(X_train.shape, Y_train.shape))
Previous X_train shape: (60000, 28, 28) Previous Y_train shape:(60000,) New X_train shape: (60000, 784) New Y_train shape:(60000, 10)
Setting up parameters¶
In [5]:
input_size = 784
batch_size = 200
hidden1 = 400
hidden2 = 20
epochs = 2
Building the FCN Model¶
In [6]:
###4.Build the model
model = Sequential()
model.add(Dense(hidden1, input_dim=input_size, activation='relu'))
# output = relu (dot (W, input) + bias)
model.add(Dense(hidden2, activation='relu'))
model.add(Dense(classes, activation='softmax'))
# Compilation
model.compile(loss='categorical_crossentropy',
metrics=['accuracy'], optimizer='sgd')
model.summary()
_________________________________________________________________ Layer (type) Output Shape Param # ================================================================= dense_1 (Dense) (None, 400) 314000 _________________________________________________________________ dense_2 (Dense) (None, 20) 8020 _________________________________________________________________ dense_3 (Dense) (None, 10) 210 ================================================================= Total params: 322,230 Trainable params: 322,230 Non-trainable params: 0 _________________________________________________________________
Training The Model¶
In [7]:
# Fitting on Data
model.fit(X_train, Y_train, batch_size=batch_size, epochs=10, verbose=2)
###5.Test
Epoch 1/10 - 12s - loss: 1.4482 - acc: 0.6251 Epoch 2/10 - 3s - loss: 0.6239 - acc: 0.8482 Epoch 3/10 - 3s - loss: 0.4582 - acc: 0.8798 Epoch 4/10 - 3s - loss: 0.3941 - acc: 0.8936 Epoch 5/10 - 3s - loss: 0.3579 - acc: 0.9011 Epoch 6/10 - 4s - loss: 0.3328 - acc: 0.9070 Epoch 7/10 - 3s - loss: 0.3138 - acc: 0.9118 Epoch 8/10 - 3s - loss: 0.2980 - acc: 0.9157 Epoch 9/10 - 3s - loss: 0.2849 - acc: 0.9191 Epoch 10/10 - 3s - loss: 0.2733 - acc: 0.9223
Out[7]:
<keras.callbacks.History at 0x272375a7240>
Testing The Model¶
In [8]:
score = model.evaluate(X_test, Y_test, verbose=1)
print('\n''Test accuracy:', score[1])
mask = range(10,20)
X_valid = X_test[mask]
y_pred = model.predict_classes(X_valid)
print(y_pred)
plt.figure(figsize=(20, 4))
for i in range(n):
# display original
ax = plt.subplot(2, n, i + 1)
plt.imshow(X_valid[i].reshape(28, 28))
plt.gray()
ax.get_xaxis().set_visible(False)
ax.get_yaxis().set_visible(False)
plt.show()
plt.close()
10000/10000 [==============================] - 1s 121us/step Test accuracy: 0.9257 [0 6 9 0 1 5 9 7 3 4]
In [ ]: