TensorFlow - Revision history

192.168.1.1: /* Using TensorFlow (without Keras) */

2017-12-06T22:29:11Z

Using TensorFlow (without Keras)

← Older revision		Revision as of 18:29, 6 December 2017
Line 111:		Line 111:
	==Using TensorFlow (without Keras)==		==Using TensorFlow (without Keras)==
	While Keras makes set-up super easy, there are some limitations.		While Keras makes set-up super easy, there are some limitations.

			...

2017-12-06T22:28:34Z

Using TensorFlow (with Keras)

← Older revision		Revision as of 18:28, 6 December 2017
Line 108:		Line 108:
	print(model.predict_proba(X)) #Display output weights		print(model.predict_proba(X)) #Display output weights
	print("Last epoch: " + str(lastEpoch)) #Display last epoch		print("Last epoch: " + str(lastEpoch)) #Display last epoch

			==Using TensorFlow (without Keras)==
			While Keras makes set-up super easy, there are some limitations.

2017-12-05T21:11:30Z

Using TensorFlow (with Keras)

← Older revision		Revision as of 17:11, 5 December 2017
Line 37:		Line 37:
	Keras is a high level API that runs on top of TensorFlow, Theano, or CNTK. It's supposed to be very user friendly and easy to use.		Keras is a high level API that runs on top of TensorFlow, Theano, or CNTK. It's supposed to be very user friendly and easy to use.
	py -m pip install keras		py -m pip install keras

			This link outlines many of the available arguements:
			https://keras.io/models/model/

	===Toy XOR Network===		===Toy XOR Network===

2017-12-05T20:57:28Z

Toy XOR Network

← Older revision		Revision as of 16:57, 5 December 2017
Line 52:		Line 52:
	#Build the Model		#Build the Model
	model = Sequential()		model = Sequential()
	model.add(Dense(8, input_dim=2)) #		model.add(Dense(8, input_dim=2, use_bias = True)) #
	model.add(Activation('tanh')) #		model.add(Activation('tanh')) #
	model.add(Dense(1)) #		model.add(Dense(1, use_bias = True)) #
	model.add(Activation('sigmoid'))		model.add(Activation('sigmoid'))

2017-12-05T20:55:59Z

Smart XOR Network

← Older revision		Revision as of 16:55, 5 December 2017
Line 73:		Line 73:
	import numpy as np		import numpy as np

	#Configure stopping point (error less than .05)		#Configure stopping point (error less than .005)
	lastEpoch = 0		lastEpoch = 0
	class EarlyStoppingByLossVal(Callback):		class EarlyStoppingByLossVal(Callback):
	def __init__(self, monitor='val_loss', value=0.05, verbose=0):		def __init__(self, monitor='val_loss', value=0.005, verbose=0):
	super(Callback, self).__init__()		super(Callback, self).__init__()
	self.monitor = monitor		self.monitor = monitor

2017-12-05T20:49:36Z

Smart XOR Network

← Older revision		Revision as of 16:49, 5 December 2017
Line 94:		Line 94:
	#Build the Model		#Build the Model
	model = Sequential()		model = Sequential()
	model.add(Dense(8, input_dim=2,use_bias = ~~False~~, kernel_initializer = VarianceScaling())) #		model.add(Dense(8, input_dim=2,use_bias = True, kernel_initializer = VarianceScaling())) #
	model.add(Activation('tanh')) #		model.add(Activation('tanh')) #
	model.add(Dense(1, use_bias = ~~False~~, kernel_initializer = VarianceScaling())) #		model.add(Dense(1, use_bias = True, kernel_initializer = VarianceScaling())) #
	model.add(Activation('sigmoid')) #Activation function		model.add(Activation('sigmoid')) #Activation function

2017-12-05T20:23:24Z

Smart XOR Network

← Older revision		Revision as of 16:23, 5 December 2017
Line 73:		Line 73:
	import numpy as np		import numpy as np

	#Configure stopping point (error less than .~~005~~)		#Configure stopping point (error less than .05)
	lastEpoch = 0		lastEpoch = 0
	class EarlyStoppingByLossVal(Callback):		class EarlyStoppingByLossVal(Callback):
	def __init__(self, monitor='val_loss', value=0.~~005~~, verbose=0):		def __init__(self, monitor='val_loss', value=0.05, verbose=0):
	super(Callback, self).__init__()		super(Callback, self).__init__()
	self.monitor = monitor		self.monitor = monitor

2017-12-05T20:19:40Z

Smart XOR Network

← Older revision		Revision as of 16:19, 5 December 2017
Line 102:		Line 102:
	model.compile(loss='mean_squared_error', optimizer=sgd) #Set error correction method (e.g. 'mean_squared_error' or 'binary_crossentropy')		model.compile(loss='mean_squared_error', optimizer=sgd) #Set error correction method (e.g. 'mean_squared_error' or 'binary_crossentropy')

	model.fit(X, y, verbose=1, batch_size=4, epochs=~~1000~~, callbacks = [EarlyStoppingByLossVal()]) #larger batches should require fewer epochs		model.fit(X, y, verbose=1, batch_size=4, epochs=10000, callbacks = [EarlyStoppingByLossVal()]) #larger batches should require fewer epochs
	print(model.predict_proba(X)) #Display output weights		print(model.predict_proba(X)) #Display output weights
	print("Last epoch: " + str(lastEpoch)) #Display last epoch		print("Last epoch: " + str(lastEpoch)) #Display last epoch

2017-12-05T20:15:59Z

Smart XOR Network

← Older revision		Revision as of 16:15, 5 December 2017
Line 73:		Line 73:
	import numpy as np		import numpy as np

	#Configure stopping point (error less than .~~008~~)		#Configure stopping point (error less than .005)
	lastEpoch = 0		lastEpoch = 0
	class EarlyStoppingByLossVal(Callback):		class EarlyStoppingByLossVal(Callback):
	def __init__(self, monitor='val_loss', value=0.~~008~~, verbose=0):		def __init__(self, monitor='val_loss', value=0.005, verbose=0):
	super(Callback, self).__init__()		super(Callback, self).__init__()
	self.monitor = monitor		self.monitor = monitor

2017-12-05T20:12:46Z

Toy XOR Network