if __name__ == '__main__': # set the random seed for reproducibility of results # setting the random seed forces the same results of randoming each # time you start the program - that way you can demonstrate your results np.random.seed(11071998) # Load the train / test data # X is the input matrix, y is the target vector # X can be a vector (and will be, in the first assignment) as well """ To change the function being approximated, just change the paths to the dataset in the arguments of the data loader.s """ X_train, y_train = dataLoader.loadFrom(SIN_TRAIN) X_test, y_test = dataLoader.loadFrom(SIN_TEST) # for check, print out the shapes of the input variables # the first dimension is the number of input samples, the second dimension # is the number of variables print "Train data shapes: ", X_train.shape, y_train.shape print "Test data shapes: ", X_test.shape, y_test.shape # The dimensionality of the input layer of the network is the second # dimension of the shape if len(X_train.shape) > 1: input_size = X_train.shape[1] else:
if __name__ == '__main__':
# set the random seed for reproducibility of results
# setting the random seed forces the same results of randoming each
# time you start the program - that way you can demonstrate your results
np.random.seed(11071998)
# Load the train / test data
# X is the input matrix, y is the target vector
# X can be a vector (and will be, in the first assignment) as well
"""
To change the function being approximated, just change the paths
to the dataset in the arguments of the data loader.s
"""
#X_train, y_train = dataLoader.loadFrom(SIN_TRAIN)
#X_test, y_test = dataLoader.loadFrom(SIN_TEST)
X_train, y_train = dataLoader.loadFrom(ROSENBROCK_TRAIN)
X_test, y_test = dataLoader.loadFrom(ROSENBROCK_TEST)
# for check, print out the shapes of the input variables
# the first dimension is the number of input samples, the second dimension
# is the number of variables
print "Train data shapes: ", X_train.shape, y_train.shape
print "Test data shapes: ", X_test.shape, y_test.shape
# The dimensionality of the input layer of the network is the second
# dimension of the shape
if len(X_train.shape) > 1:
input_size = X_train.shape[1]
else:
# Load the train / test data # X is the input matrix, y is the target vector # X can be a vector (and will be, in the first assignment) as well """ To change the function being approximated, just change the paths to the dataset in the arguments of the data loader.s """ sinusoida = False rastrigin = True rosenbrock = False if sinusoida: X_train, y_train = dataLoader.loadFrom(SIN_TRAIN) X_test, y_test = dataLoader.loadFrom(SIN_TEST) if rastrigin: X_train, y_train = dataLoader.loadFrom(RASTRIGIN_TRAIN) X_test, y_test = dataLoader.loadFrom(RASTRIGIN_TEST) if rosenbrock: X_train, y_train = dataLoader.loadFrom(ROSENBROCK_TRAIN) X_test, y_test = dataLoader.loadFrom(ROSENBROCK_TEST) # for check, print out the shapes of the input variables # the first dimension is the number of input samples, the second dimension # is the number of variables print "Train data shapes: ", X_train.shape, y_train.shape