def classify_XOR():
#Load Dataset
P, T, Ptest, Ttest = dts.loadDataset_XOR()
input_shape = (P.shape[1], )
output_shape = T.shape[1] # Number of Dense neurons at output layer
### Build Model
dendral_neurons = 6
lr = 0.08971484393708822
activation = 'tanh'
model = bm.build_HybridModel_MLNN(dendral_neurons, activation, input_shape,
output_shape)
[hist, train_time] = bm.train_HybridModel_MLNN(model,
lr,
P,
T,
Ptest,
Ttest,
batch_size=512,
nb_epoch=100,
v_verbose=False)
print("\n\t Dataset XOR: ")
print("\n\t Classificacion: " + str(hist.history['val_acc'][-1]))
plt_util.my_plot_train_loss(hist)
def classify_2C_5L_Spiral():
#Load Dataset
P, T, Ptest, Ttest = dts.loadDataset_Espiral_2Class_N_Loops()
input_shape = (P.shape[1], )
output_shape = T.shape[1] # Number of Dense neurons at output layer
### Build Model
dendral_neurons = 250
lr = 0.2
activation = 'tanh'
batch_size = 512
model = bm.build_HybridModel_MLNN(dendral_neurons, activation, input_shape,
output_shape)
[hist, train_time] = bm.train_HybridModel_MLNN(model,
lr,
P,
T,
Ptest,
Ttest,
batch_size=batch_size,
nb_epoch=1000,
v_verbose=False)
print("\n\t Dataset 2 class 5 Loops spiral : ")
print("\n\t Classificacion: " + str(hist.history['val_acc'][-1]))
plt_util.my_plot_train_loss(hist)
plt_util.plot_decision_boundary_2_class(P,
model,
batch_size,
h=0.05,
half_dataset=True,
expand=0.5,
x_lim=45,
y_lim=45)