def validateTitanic():
# Data
X, Y = getDataValidation()
# Network
inputSize = 7
layerCount = 2
networkConf = nnConf.NeuralNetworkConf(inputSize, layerCount)
networkConf.layerConf[0].neuronCount = 20
networkConf.layerConf[0].activationFn = "relu"
networkConf.layerConf[0].weightInitializerMethod = "random"
networkConf.layerConf[1].neuronCount = 1
networkConf.layerConf[1].activationFn = "sigmoid"
networkConf.layerConf[1].weightInitializerMethod = "random"
networkConf.Lambda = .00009 # 0.0001
networkConf.maxIter = 500
accuracyList = nnUtils.validateNN(X, Y, networkConf, 5, showLearning=False)
print(accuracyList)
print("Mean Accuracy: ", np.mean(accuracyList))
# Store Neural network settings and state
# print(NN.getGlobalConf())
# print(networkConf)
nnUtils.saveConfig(networkConf, nn.getGlobalConf())
nnUtils.restoreConfig()
accuracyList = []
return (accuracyList)
def runAndMeasure():
# Data
X, Y = getDataValidation()
# Network
inputSize = 7
layerCount = 2
netConf = nnConf.NeuralNetworkConf(inputSize, layerCount)
netConf.layerConf[0].neuronCount = 20
netConf.layerConf[0].activationFn = "relu"
netConf.layerConf[0].weightInitializerMethod = "random"
netConf.layerConf[1].neuronCount = 1
netConf.layerConf[1].activationFn = "sigmoid"
netConf.layerConf[1].weightInitializerMethod = "random"
netConf.Lambda = .00009 # 0.0001
netConf.maxIter = 500
accuracyList = nnUtils.validateNN(X, Y, netConf, 5,
showLearning=False)
netConf.accuracy = np.mean(accuracyList)
print(accuracyList)
print("Mean Accuracy: ", netConf.accuracy)
return (netConf, nn.getGlobalConf())
def loadAndValidate(trainX, trainY):
netConf, wrightsBias = nnUtils.restoreConfig("nn.json")
accuracyList = nnUtils.validateNN(trainX,
trainY,
netConf,
5,
showLearning=False,
wtReuse=True,
wtAndBias=wrightsBias)
print("Best accuracy on validation: ", np.mean(accuracyList))
def loadAndRun():
# Data
X, Y = getDataValidation()
netConf, wrightsBias = nnUtils.restoreConfig("nn.json")
accuracyList = nnUtils.validateNN(X, Y, netConf, 5,
showLearning=False,
wtReuse=True,
wtAndBias=wrightsBias)
print(accuracyList)
print("Mean Accuracy: ", np.mean(accuracyList))
def findAndStoreBestInitialWeights(X, Y, netConf, noOfPass):
bestAccuracy = 0
n_splits = 5
for i in range(noOfPass):
accuracyList = nnUtils.validateNN(X,
Y,
netConf,
n_splits,
showLearning=False)
if np.mean(accuracyList) > bestAccuracy:
bestAccuracy = np.mean(accuracyList)
# Store Neural network settings and state
print("Got Better accuracy: ", bestAccuracy)
netConf.accuracy = bestAccuracy
nnUtils.saveConfig(netConf, nn.getGlobalConf(), "nn.json")
def getBestMaxIter(X, Y, netConf):
bestMaxIter = 0
bestAccuracy = 0
for MaxIter in range(100, 1000, 100):
netConf.maxIter = MaxIter
n_splits = 5
accuracyList = nnUtils.validateNN(X,
Y,
netConf,
n_splits,
showLearning=False,
wtReuse=True,
wtAndBias=nn.getGlobalConf())
if np.mean(accuracyList) > bestAccuracy:
bestAccuracy = np.mean(accuracyList)
bestMaxIter = MaxIter
return (bestMaxIter, bestAccuracy)
def getBestLambda(X, Y, netConf):
bestLambda = 0
bestAccuracy = 0
for Lambda in np.linspace(0.00001, 0.0001, 200):
netConf.Lambda = Lambda
n_splits = 5
accuracyList = nnUtils.validateNN(X,
Y,
netConf,
n_splits,
showLearning=False,
wtReuse=True,
wtAndBias=nn.getGlobalConf())
if np.mean(accuracyList) > bestAccuracy:
bestAccuracy = np.mean(accuracyList)
bestLambda = Lambda
return (bestLambda, bestAccuracy)
def getBestNeuronCount(X, Y, netConf):
bestNeuronCount = 0
bestAccuracy = 0
for neuronCount in range(2, 30):
netConf.layerConf[0].neuronCount = neuronCount
n_splits = 5
accuracyList = nnUtils.validateNN(X,
Y,
netConf,
n_splits,
showLearning=False,
wtReuse=True,
wtAndBias=nn.getGlobalConf())
if np.mean(accuracyList) > bestAccuracy:
bestAccuracy = np.mean(accuracyList)
bestNeuronCount = neuronCount
return (bestNeuronCount, bestAccuracy)
if __name__ == "__main__":
# 1. Parse data and get the useful features
trainX, trainY = getDataValidation()
# 2. Get the best initial network configuration.
# * Decide and fix network depth(getNetConf).
netConf = getInitialNetConf()
# * Run evaluation once with randomized initial weight and store
# the weight(Evaluate and store function, takes a flag for
# random initialization).
n_splits = 5
accuracyList = nnUtils.validateNN(trainX,
trainY,
netConf,
n_splits,
showLearning=False)
print("Initial Mean Accuracy: ", np.mean(accuracyList))
# * Run evaluation varying different network parameters with
# the fixed initial weights.
# netConf.Lambda, accuracy = getBestLambda(trainX, trainY, netConf)
# netConf.maxIter, accuracy = getBestMaxIter(trainX, trainY, netConf)
netConf.Lambda = 0.0000946
netConf.maxIter = 500
print("Best Lambda: ", netConf.Lambda)
print("Best MaxIter: ", netConf.maxIter)
print("Best Config Mean Accuracy: ", np.mean(accuracyList))
# 3. With the best network configuration get best set of Weights.
