Keras hello world

Keras hello world.md

Using Keras (now part of TensorFlow) is really easy. The complexity comes when you deal with large amounts of data figuring out the topology of a neural network. With the topology comes hyperparameter tuning and all that. It's a bit like painting: it's easy to hold a brush but it takes years to paint something worth looking at.

	import tensorflow as tf
	from tensorflow.keras.models import Sequential
	from tensorflow.keras.layers import Dense


	import numpy as np
	x_input = np.array([[1,2,3,4,5]])
	y_input = np.array([[10]])


	model = Sequential()
	model.add(Dense(units=32, activation="tanh", input_dim=x_input.shape[1], kernel_initializer='random_normal'))
	model.add(Dense(units=1, kernel_initializer='random_normal'))


	model.compile(loss='mse', optimizer='sgd', metrics=['accuracy'])


	model.summary()

_________________________________________________________________
Layer (type)                 Output Shape              Param #
=================================================================
dense_10 (Dense)             (None, 32)                192
_________________________________________________________________
dense_11 (Dense)             (None, 1)                 33
=================================================================
Total params: 225
Trainable params: 225
Non-trainable params: 0
_________________________________________________________________



	history = model.fit(x_input, y_input, epochs=10, batch_size=32)

Epoch 1/10
1/1 [==============================] - 0s 163ms/step - loss: 36.1381 - acc: 0.0000e+00
Epoch 2/10
1/1 [==============================] - 0s 1ms/step - loss: 0.0645 - acc: 1.0000
Epoch 3/10
1/1 [==============================] - 0s 2ms/step - loss: 0.0075 - acc: 1.0000
Epoch 4/10
1/1 [==============================] - 0s 961us/step - loss: 8.8381e-04 - acc: 1.0000
Epoch 5/10
1/1 [==============================] - 0s 1ms/step - loss: 1.0349e-04 - acc: 1.0000
Epoch 6/10
1/1 [==============================] - 0s 1ms/step - loss: 1.2137e-05 - acc: 1.0000
Epoch 7/10
1/1 [==============================] - 0s 882us/step - loss: 1.4188e-06 - acc: 1.0000
Epoch 8/10
1/1 [==============================] - 0s 2ms/step - loss: 1.6660e-07 - acc: 1.0000
Epoch 9/10
1/1 [==============================] - 0s 1ms/step - loss: 1.8859e-08 - acc: 1.0000
Epoch 10/10
1/1 [==============================] - 0s 1ms/step - loss: 2.2737e-09 - acc: 1.0000



	model.predict(x_input, batch_size=128)

array([[10.000018]], dtype=float32)

	model.predict(np.array([[1,2,5,4,5]]))

array([[7.1825438]], dtype=float32)