def test():
# TODO : Test Later
print('==> Testing network..')
# Make predictions on full X_test mels
y_predicted = accuracy.predict_class_all(create_segmented_mels(X_test), a_net)
# Print statistics
print(np.sum(accuracy.confusion_matrix(y_predicted, y_test),axis=1))
print(accuracy.confusion_matrix(y_predicted, y_test))
print(accuracy.get_accuracy(y_predicted,y_test))
X_test = pool.map(get_wav, X_test)
# Convert to MFCC
if DEBUG:
print('converting to mfcc')
X_train = pool.map(to_mfcc, X_train)
X_test = pool.map(to_mfcc, X_test)
# Create segments from MFCCs
X_train, y_train = make_segments(X_train, y_train)
X_validation, y_validation = make_segments(X_test, y_test)
# Randomize training segments
X_train, _, y_train, _ = train_test_split(X_train, y_train, test_size=0)
# Train model
model = train_model(np.array(X_train), np.array(y_train), np.array(X_validation),np.array(y_validation))
# Make predictions on full X_test MFCCs
y_predicted = accuracy.predict_class_all(create_segmented_mfccs(X_test), model)
# Print statistics
print train_count
print test_count
print acc_to_beat
print np.sum(accuracy.confusion_matrix(y_predicted, y_test),axis=1)
print accuracy.confusion_matrix(y_predicted, y_test)
print accuracy.get_accuracy(y_predicted,y_test)
# Save model
save_model(model, model_filename)
if DEBUG:
print('Converting to MFCC....')
X_train = pool.map(to_mfcc, X_train)
X_test = pool.map(to_mfcc, X_test)
# Create segments from MFCCs
X_train, y_train = make_segments(X_train, y_train)
X_validation, y_validation = make_segments(X_test, y_test)
# Randomize training segments
X_train, _, y_train, _ = train_test_split(X_train, y_train, test_size=0.2)
# Train model
model = train_model(np.array(X_train), np.array(y_train), np.array(X_validation),np.array(y_validation))
# Make predictions on full X_test MFCCs
y_predicted = accuracy.predict_class_all(create_segmented_mfccs(X_test), model)
# Save model
save_model(model, model_filename)
# Print statistics
print('Training samples:', train_count)
print('Testing samples:', test_count)
print('Accuracy to beat:', acc_to_beat)
print('Confusion matrix of total samples:\n', np.sum(accuracy.confusion_matrix(y_predicted, y_test),axis=1))
print('Confusion matrix:\n',accuracy.confusion_matrix(y_predicted, y_test))
print('Accuracy:', accuracy.get_accuracy(y_predicted,y_test))
zs = var_auto.encode(data)
model = KMeans(n_clusters=n_clusters, random_state=42, n_init=20)
#model = SpectralClustering(n_clusters=n_clusters, eigen_solver='arpack', affinity="nearest_neighbors")
#model = DBSCAN(eps=.2)
results = model.fit_predict(zs)
#joblib.dump(model, "F:/Fer-novo/Diplomski rad/modeli/saves/kmeans.pkl") # optional storing the model for future use
# tsne on clusters on train data
tsne = TSNE(n_components = 2, random_state=42)
tsne_results = tsne.fit_transform(zs[-1300:])
P.scatterTsne(tsne_results, results[-1300:])
#tsne for test data
test_data_all = np.array([]).reshape(0, num)
for test_data in test_set_list:
test_data_all = np.append(test_data_all, test_data, axis=0)
padding = np.zeros(1000*num).reshape(1000, num)
test_data_all = np.append(test_data_all, padding.reshape(-1, num), axis=0)
zs_test = var_auto.encode(test_data_all)[:(test_set_size * len(test_set_list))]
tsne_results_test = tsne.fit_transform(zs_test)
P.scatterTsne(tsne_results_test, model.predict(zs_test))
#save results
#D.saveResults(save_folder, results, original_data)
print(A.confusion_matrix(test_set_list, var_auto, model, n_clusters, num=num, test_set_size=test_set_size))
# X_validation, y_validation = make_segments(X_test, y_test)
X_train, X_validation, y_train, y_validation = train_test_split(
X_train, y_train, test_size=0.15)
# print "Validation shape: {}".format(X_validation)
# Randomize training segments
X_train, _, y_train, _ = train_test_split(X_train, y_train, test_size=0)
if network == 'cnn':
# Train model
model = train_model(np.array(X_train), np.array(y_train),
np.array(X_validation), np.array(y_validation),
EPOCHS)
# Make predictions on full X_test MFCCs
y_predicted = accuracy.predict_class_all(
create_segmented_mfccs(X_test), model, 'cnn')
class_sum = np.sum(accuracy.confusion_matrix(y_predicted, y_test),
axis=1)
confusion_matrix = accuracy.confusion_matrix(y_predicted, y_test)
print confusion_matrix
print accuracy.get_accuracy(y_predicted, y_test)
show_confusion_matrix(confusion_matrix, plt,
['mandarin', 'arabic', 'english'], 'cnn')
if network == 'lstm':
# Train Lstm Model
lstm = train_lstm_model(np.array(X_train), np.array(y_train),
np.array(X_validation), np.array(y_validation),
EPOCHS)
y_predicted_lstm = accuracy.predict_class_all(
create_segmented_mfccs(X_test), lstm, 'lstm')
print np.sum(accuracy.confusion_matrix(y_predicted_lstm, y_test),
# print (trainer)
# Train model
model = train_model(np.array(X_train), np.array(y_train),
np.array(X_validation), np.array(y_validation))
# model = load_model("model.h5")
# predicted = model.predict (k.create_segmented_mfccs(X_test))
# for i in predicted:
# print (i)
# Make predictions on full X_test MFCCs
y_predicted = accuracy.predict_class_all(
k.create_segmented_mfccs(X_test), model)
# print (y_predicted)
# for i in y_predicted:
# print (i)
# Print statistics
print('Training samples:', train_count)
print('Testing samples:', test_count)
print('Accuracy to beat:', acc_to_beat)
print('Confusion matrix of total samples:\n',
np.sum(accuracy.confusion_matrix(y_predicted, y_test), axis=1))
print('Confusion matrix:\n',
accuracy.confusion_matrix(y_predicted, y_test))
print('Accuracy:', accuracy.get_accuracy(y_predicted, y_test))
results.append(accuracy.confusion_matrix(y_predicted, y_test))
acc.append(accuracy.get_accuracy(y_predicted, y_test))
# Save model
print(results)
print(acc)
save_model(model, model_filename)