def text_classification_query(self,
instruction,
drop=None,
preprocess=True,
label_column=None,
test_size=0.2,
random_state=49,
learning_rate=1e-2,
epochs=20,
monitor="val_loss",
batch_size=32,
max_text_length=200,
max_features=20000,
generate_plots=True,
save_model=False,
save_path=os.getcwd()):
"""
function to apply text_classification algorithm for sentiment analysis
:param many params: used to hyperparametrize the function.
:return a dictionary object with all of the information for the algorithm.
"""
if test_size < 0:
raise Exception("Test size must be a float between 0 and 1")
if test_size >= 1:
raise Exception(
"Test size must be a float between 0 and 1 (a test size greater than or equal to 1 results in no training "
"data)")
if epochs < 1:
raise Exception(
"Epoch number is less than 1 (model will not be trained)")
if batch_size < 1:
raise Exception("Batch size must be equal to or greater than 1")
if max_text_length < 1:
raise Exception("Max text length must be equal to or greater than 1")
if save_model:
if not os.path.exists(save_path):
raise Exception("Save path does not exists")
if test_size == 0:
testing = False
else:
testing = True
data = DataReader(self.dataset)
data = data.data_generator()
if preprocess:
data.fillna(0, inplace=True)
if drop is not None:
data.drop(drop, axis=1, inplace=True)
if label_column is None:
label = "label"
else:
label = label_column
X, Y, target = get_target_values(data, instruction, label)
Y = np.array(Y)
classes = np.unique(Y)
logger("->", "Target Column Found: {}".format(target))
vocab = {}
if preprocess:
logger("Preprocessing data")
X = lemmatize_text(text_clean_up(X.array))
vocab = X
X = encode_text(X, X)
X = np.array(X)
model = get_keras_text_class(max_features, len(classes), learning_rate)
logger("Building Keras LSTM model dynamically")
X_train, X_test, y_train, y_test = train_test_split(
X, Y, test_size=test_size, random_state=random_state)
X_train = sequence.pad_sequences(X_train, maxlen=max_text_length)
X_test = sequence.pad_sequences(X_test, maxlen=max_text_length)
y_vals = np.unique(np.append(y_train, y_test))
label_mappings = {}
for i in range(len(y_vals)):
label_mappings[y_vals[i]] = i
map_func = np.vectorize(lambda x: label_mappings[x])
y_train = map_func(y_train)
y_test = map_func(y_test)
logger("Training initial model")
# early stopping callback
es = EarlyStopping(monitor=monitor, mode='auto', verbose=0, patience=5)
history = model.fit(X_train,
y_train,
validation_data=(X_test, y_test),
batch_size=batch_size,
epochs=epochs,
callbacks=[es],
verbose=0)
logger(
"->", "Final training loss: {}".format(
history.history["loss"][len(history.history["loss"]) - 1]))
if testing:
logger(
"->", "Final validation loss: {}".format(
history.history["val_loss"][len(history.history["val_loss"]) -
1]))
logger(
"->", "Final validation accuracy: {}".format(
history.history["val_accuracy"][
len(history.history["val_accuracy"]) - 1]))
losses = {
'training_loss': history.history['loss'],
'val_loss': history.history['val_loss']
}
accuracy = {
'training_accuracy': history.history['accuracy'],
'validation_accuracy': history.history['val_accuracy']
}
else:
logger("->",
"Final validation loss: {}".format("0, No validation done"))
losses = {'training_loss': history.history['loss']}
accuracy = {'training_accuracy': history.history['accuracy']}
plots = {}
if generate_plots:
# generates appropriate classification plots by feeding all
# information
logger("Generating plots")
plots = generate_classification_plots(history, X, Y, model, X_test,
y_test)
if save_model:
save(model, save_model, save_path=save_path)
logger(
"Storing information in client object under key 'text_classification'")
# storing values the model dictionary
self.models["text_classification"] = {
"model": model,
"classes": classes,
"plots": plots,
"target": Y,
"vocabulary": vocab,
"interpreter": label_mappings,
"max_text_length": max_text_length,
'test_data': {
'X': X_test,
'y': y_test
},
'losses': losses,
'accuracy': accuracy
}
clearLog()
return self.models["text_classification"]
def tune_helper(model_to_tune=None,
dataset=None,
models=None,
max_layers=10,
min_layers=2,
min_dense=32,
max_dense=512,
executions_per_trial=3,
max_trials=1,
activation='relu',
loss='categorical_crossentropy',
metrics='accuracy',
seed=42,
objective='val_accuracy',
generate_plots=True,
directory='my_dir',
epochs=10,
step=32,
patience=1,
verbose=0,
test_size=0.2):
'''
Helper function that calls the appropriate tuning function
:param instruction: the objective that you want to reduce dimensions to maximize
:return the updated models dictionary
'''
print("")
logger("Getting target model for tuning...")
# checks to see which requested model is in the self.models
# processing for regression feed forward NN
if model_to_tune == 'regression_ANN':
logger("Reading in data")
logger("Tuning model hyperparameters...")
dataReader = DataReader(dataset)
data = dataReader.data_generator()
target = models['regression_ANN']['target']
target_column = data[models['regression_ANN']['target']]
data = models['regression_ANN']['preprocesser'].transform(
data.drop(target, axis=1))
returned_model, returned_pms, history, X_test, y_test = tuneReg(
data.values,
target_column.values,
max_layers=max_layers,
min_layers=min_layers,
min_dense=min_dense,
max_dense=max_dense,
executions_per_trial=executions_per_trial,
max_trials=max_trials,
epochs=epochs,
activation=activation,
step=step,
directory=directory,
verbose=verbose,
test_size=test_size)
plots = {}
logger("->",
'Best Hyperparameters Found: {}'.format(returned_pms.values))
if generate_plots:
logger("Generating updated plots")
init_plots, plot_names = generate_regression_plots(
history, data, target_column)
for x in range(len(plot_names)):
plots[str(plot_names[x])] = init_plots[x]
models['regression_ANN'] = {
'id': models['regression_ANN']['id'],
'model': returned_model,
'target': target,
"plots": plots,
'preprocesser': models['regression_ANN']['preprocesser'],
'interpreter': models['regression_ANN']['interpreter'],
'test_data': {
'X': X_test,
'y': y_test
},
'hyperparameters': returned_pms.values,
'losses': {
'training_loss': history.history['loss'],
'val_loss': history.history['val_loss']
}
}
logger("Re-stored model under 'regression_ANN' key")
# processing for classification feed forward NN
elif model_to_tune == "classification_ANN":
logger("Reading in data")
logger("Tuning model hyperparameters...")
dataReader = DataReader(dataset)
data = dataReader.data_generator()
target = models['classification_ANN']['target']
target_column = data[models['classification_ANN']['target']]
data = models['classification_ANN']['preprocesser'].transform(
data.drop(target, axis=1))
returned_model, returned_pms, history, X_test, y_test = tuneClass(
data,
target_column,
models['classification_ANN']['num_classes'],
max_layers=max_layers,
min_layers=min_layers,
min_dense=min_dense,
max_dense=max_dense,
executions_per_trial=executions_per_trial,
max_trials=max_trials,
activation=activation,
loss=loss,
directory=directory,
metrics=metrics,
epochs=epochs,
step=step,
verbose=verbose,
test_size=test_size)
plots = {}
logger("->",
'Best Hyperparameters Found: {}'.format(returned_pms.values))
if generate_plots:
logger("Generating updated plots")
plots = generate_classification_plots(history, data, target_column,
returned_model, X_test,
y_test)
logger("Re-stored model under 'classification_ANN' key")
models['classification_ANN'] = {
'id': models['classification_ANN']['id'],
'model': returned_model,
'hyperparameters': returned_pms.values,
'plots': plots,
'preprocesser': models['classification_ANN']['preprocesser'],
'interpreter': models['classification_ANN']['interpreter'],
'test_data': {
'X': X_test,
'y': y_test
},
'target': target,
'losses': {
'training_loss': history.history['loss'],
'val_loss': history.history['val_loss']
},
'accuracy': {
'training_accuracy': history.history['accuracy'],
'validation_accuracy': history.history['val_accuracy']
}
}
elif model_to_tune == "convolutional_NN":
logger("Tuning model hyperparameters...")
X_train, X_test, height, width, num_classes = get_image_data(models)
logger('Located image data')
model, returned_pms, history = tuneCNN(
X_train,
X_test,
height,
width,
num_classes,
executions_per_trial=executions_per_trial,
max_trials=max_trials,
seed=seed,
objective=objective,
directory=directory,
patience=patience,
epochs=epochs,
verbose=verbose,
test_size=test_size)
logger("->", "Optimal image size identified: {}".format(
(height, width, 3)))
logger('Packaging HyperModel')
logger("->",
'Best Hyperparameters Found: {}'.format(returned_pms.values))
logger("Re-stored model under 'convolutional_NN' key")
models['convolutional_NN'] = {
'id': models['convolutional_NN']['id'],
'data_type': models['convolutional_NN']['data_type'],
'data_path': models['convolutional_NN']['data_path'],
'data': {
'train': X_train,
'test': X_test
},
'shape': models['convolutional_NN']['shape'],
'model': model,
'num_classes': models['convolutional_NN']['num_classes'],
'data_sizes': models['convolutional_NN']['data_sizes'],
'losses': {
'training_loss': history.history['loss'],
'val_loss': history.history['val_loss']
},
'accuracy': {
'training_accuracy': history.history['accuracy'],
'validation_accuracy': history.history['val_accuracy']
}
}
clearLog()
return models
def classification_ann(instruction,
callback=False,
dataset=None,
text=[],
ca_threshold=None,
preprocess=True,
callback_mode='min',
drop=None,
random_state=49,
test_size=0.2,
epochs=50,
generate_plots=True,
maximizer="val_accuracy",
save_model=False,
save_path=os.getcwd(),
add_layer={}):
'''
Body of the classification function used that is called in the neural network query
if the data is categorical.
:param many parameters: used to preprocess, tune, plot generation, and parameterizing the neural network trained.
:return dictionary that holds all the information for the finished model.
'''
if dataset is None:
dataReader = DataReader(get_file())
else:
dataReader = DataReader(dataset)
logger("Reading in dataset")
data = dataReader.data_generator()
if drop is not None:
data.drop(drop, axis=1, inplace=True)
data, y, remove, full_pipeline = initial_preprocessor(
data,
instruction,
preprocess,
ca_threshold,
text,
test_size=test_size,
random_state=random_state)
logger("->", "Target column found: {}".format(remove))
# Needed to make a custom label encoder due to train test split changes
# Can still be inverse transformed, just a bit of extra work
y = pd.concat([y['train'], y['test']], axis=0)
num_classes = len(np.unique(y))
if num_classes < 2:
raise Exception("Number of classes must be greater than or equal to 2")
X_train = data['train']
X_test = data['test']
if num_classes >= 2:
# ANN needs target one hot encoded for classification
one_hotencoder = OneHotEncoder()
y = pd.DataFrame(one_hotencoder.fit_transform(
np.reshape(y.values, (-1, 1))).toarray(),
columns=one_hotencoder.get_feature_names())
y_train = y.iloc[:len(X_train)]
y_test = y.iloc[len(X_train):]
models = []
losses = []
accuracies = []
model_data = []
logger("Establishing callback function")
# early stopping callback
es = EarlyStopping(monitor=maximizer, mode='max', verbose=0, patience=5)
callback_value = None
if callback is not False:
callback_value = [es]
i = 0
model = get_keras_model_class(data, i, num_classes, add_layer)
logger("Training initial model")
history = model.fit(X_train,
y_train,
callbacks=callback_value,
epochs=epochs,
validation_data=(X_test, y_test),
verbose=0)
model_data.append(model)
models.append(history)
col_name = [[
"Initial number of layers ", "| Training Accuracy ", "| Test Accuracy "
]]
col_width = max(len(word) for row in col_name for word in row) + 2
for row in col_name:
print((" " * 2 * counter) + "| " +
("".join(word.ljust(col_width) for word in row)) + " |")
values = []
values.append(str(len(model.layers)))
values.append("| " + str(history.history['accuracy'][
len(history.history['val_accuracy']) - 1]))
values.append("| " + str(history.history['val_accuracy'][
len(history.history['val_accuracy']) - 1]))
datax = []
datax.append(values)
for row in datax:
print((" " * 2 * counter) + "| " +
("".join(word.ljust(col_width) for word in row)) + " |")
# print((" " * 2 * counter)+ tabulate(datax, headers=col_name, tablefmt='orgtbl'))
losses.append(history.history[maximizer][len(history.history[maximizer]) -
1])
accuracies.append(
history.history['val_accuracy'][len(history.history['val_accuracy']) -
1])
# keeps running model and fit functions until the validation loss stops
# decreasing
logger("Testing number of layers")
col_name = [[
"Current number of layers", "| Training Accuracy", "| Test Accuracy"
]]
col_width = max(len(word) for row in col_name for word in row) + 2
for row in col_name:
print((" " * 2 * counter) + "| " +
("".join(word.ljust(col_width) for word in row)) + " |")
datax = []
# while all(x < y for x, y in zip(accuracies, accuracies[1:])):
while (len(accuracies) <= 2 or
accuracies[len(accuracies) - 1] > accuracies[len(accuracies) - 2]):
model = get_keras_model_class(data, i, num_classes, add_layer)
history = model.fit(X_train,
y_train,
callbacks=callback_value,
epochs=epochs,
validation_data=(X_test, y_test),
verbose=0)
values = []
datax = []
values.append(str(len(model.layers)))
values.append("| " + str(history.history['accuracy'][
len(history.history['accuracy']) - 1]))
values.append("| " + str(history.history['val_accuracy'][
len(history.history['val_accuracy']) - 1]))
datax.append(values)
for row in datax:
print((" " * 2 * counter) + "| " +
("".join(word.ljust(col_width) for word in row)) + " |")
del values, datax
losses.append(
history.history[maximizer][len(history.history[maximizer]) - 1])
accuracies.append(history.history['val_accuracy'][
len(history.history['val_accuracy']) - 1])
models.append(history)
model_data.append(model)
i += 1
# print((" " * 2 * counter)+ tabulate(datax, headers=col_name, tablefmt='orgtbl'))
# del values, datax
final_model = model_data[accuracies.index(max(accuracies))]
final_hist = models[accuracies.index(max(accuracies))]
print("")
logger('->',
"Best number of layers found: " + str(len(final_model.layers)))
logger(
'->', "Training Accuracy: " + str(final_hist.history['accuracy'][
len(final_hist.history['val_accuracy']) - 1]))
logger(
'->', "Test Accuracy: " + str(final_hist.history['val_accuracy'][
len(final_hist.history['val_accuracy']) - 1]))
# genreates appropriate classification plots by feeding all information
plots = {}
if generate_plots:
plots = generate_classification_plots(models[len(models) - 1])
if save_model:
save(final_model, save_model, save_path)
print("")
logger("Stored model under 'classification_ANN' key")
clearLog()
K.clear_session()
# stores the values and plots into the object dictionary
return {
'id': generate_id(),
"model": final_model,
'num_classes': num_classes,
"plots": plots,
"target": remove,
"preprocessor": full_pipeline,
"interpreter": one_hotencoder,
'test_data': {
'X': X_test,
'y': y_test
},
'losses': {
'training_loss': final_hist.history['loss'],
'val_loss': final_hist.history['val_loss']
},
'accuracy': {
'training_accuracy': final_hist.history['accuracy'],
'validation_accuracy': final_hist.history['val_accuracy']
}
}
def convolutional(instruction=None,
read_mode=None,
preprocess=True,
data_path=None,
verbose=0,
new_folders=True,
image_column=None,
training_ratio=0.8,
fine_tune=False,
augmentation=True,
custom_arch=None,
pretrained=None,
epochs=10,
height=None,
width=None,
save_as_tfjs=None,
save_as_tflite=None,
generate_plots=True):
'''
Body of the convolutional function used that is called in the neural network query
if the data is presented in images.
:param many parameters: used to preprocess, tune, plot generation, and parameterizing the convolutional neural network trained.
:return dictionary that holds all the information for the finished model.
'''
# data_path = get_folder_dir()
logger("Generating datasets for classes")
LR = 0.001
plots = {}
if pretrained:
if not height:
height = 224
if not width:
width = 224
if height != 224 or width != 224:
raise ValueError(
"For pretrained models, both 'height' and 'width' must be 224."
)
if preprocess:
if custom_arch:
raise ValueError(
"If 'custom_arch' is not None, 'preprocess' must be set to false."
)
read_mode_info = set_distinguisher(data_path, read_mode)
read_mode = read_mode_info["read_mode"]
training_path = "/proc_training_set"
testing_path = "/proc_testing_set"
if read_mode == "setwise":
processInfo = setwise_preprocessing(data_path, new_folders, height,
width)
if not new_folders:
training_path = "/training_set"
testing_path = "/testing_set"
# if image dataset in form of csv
elif read_mode == "csvwise":
if training_ratio <= 0 or training_ratio >= 1:
raise BaseException(f"Test ratio must be between 0 and 1.")
processInfo = csv_preprocessing(read_mode_info["csv_path"],
data_path, instruction,
image_column, training_ratio,
height, width)
# if image dataset in form of one folder containing class folders
elif read_mode == "classwise":
if training_ratio <= 0 or training_ratio >= 1:
raise BaseException(f"Test ratio must be between 0 and 1.")
processInfo = classwise_preprocessing(data_path, training_ratio,
height, width)
else:
training_path = "/training_set"
testing_path = "/testing_set"
processInfo = already_processed(data_path)
num_channels = 3
color_mode = 'rgb'
if processInfo["gray_scale"]:
num_channels = 1
color_mode = 'grayscale'
input_shape = (processInfo["height"], processInfo["width"], num_channels)
input_single = (processInfo["height"], processInfo["width"])
num_classes = processInfo["num_categories"]
loss_func = ""
output_layer_activation = ""
if num_classes > 2:
loss_func = "categorical_crossentropy"
output_layer_activation = "softmax"
elif num_classes == 2:
num_classes = 1
loss_func = "binary_crossentropy"
output_layer_activation = "sigmoid"
logger("Creating convolutional neural network dynamically")
# Convolutional Neural Network
# Build model based on custom_arch configuration if given
if custom_arch:
with open(custom_arch, "r") as f:
custom_arch_dict = json.load(f)
custom_arch_json_string = json.dumps(custom_arch_dict)
model = model_from_json(custom_arch_json_string)
# Build an existing state-of-the-art model
elif pretrained:
arch_lower = pretrained.get('arch').lower()
# If user specifies value of pretrained['weights'] as 'imagenet', weights pretrained on ImageNet will be used
if 'weights' in pretrained and pretrained.get('weights') == 'imagenet':
# Load ImageNet pretrained weights
if arch_lower == "vggnet16":
base_model = VGG16(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = Flatten()(base_model.output)
x = Dense(4096)(x)
x = Dropout(0.5)(x)
x = Dense(4096)(x)
x = Dropout(0.5)(x)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
elif arch_lower == "vggnet19":
base_model = VGG19(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = Flatten()(base_model.output)
x = Dense(4096)(x)
x = Dropout(0.5)(x)
x = Dense(4096)(x)
x = Dropout(0.5)(x)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
elif arch_lower == "resnet50":
base_model = ResNet50(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = Flatten()(base_model.output)
x = GlobalAveragePooling2D()(base_model.output)
x = Dropout(0.5)(x)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
elif arch_lower == "resnet101":
base_model = ResNet101(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = GlobalAveragePooling2D()(base_model.output)
x = Dropout(0.5)(x)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
elif arch_lower == "resnet152":
base_model = ResNet152(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = GlobalAveragePooling2D()(base_model.output)
x = Dropout(0.5)(x)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
elif arch_lower == "mobilenet":
base_model = MobileNet(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = fine_tuned_model(base_model)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
elif arch_lower == "mobilenetv2":
base_model = MobileNetV2(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = fine_tuned_model(base_model)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
elif arch_lower == "densenet121":
base_model = DenseNet121(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = fine_tuned_model(base_model)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
elif arch_lower == "densenet169":
base_model = DenseNet169(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = fine_tuned_model(base_model)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
elif arch_lower == "densenet201":
base_model = DenseNet201(include_top=False,
weights='imagenet',
input_shape=input_shape)
x = fine_tuned_model(base_model)
pred = Dense(num_classes,
activation=output_layer_activation)(x)
model = Model(base_model.input, pred)
else:
raise ModuleNotFoundError("arch \'" + pretrained.get('arch') +
"\' not supported.")
else:
# Randomly initialized weights
if arch_lower == "vggnet16":
model = VGG16(include_top=True,
weights=None,
classes=num_classes,
classifier_activation=output_layer_activation)
elif arch_lower == "vggnet19":
model = VGG19(include_top=True,
weights=None,
classes=num_classes,
classifier_activation=output_layer_activation)
elif arch_lower == "resnet50":
model = ResNet50(include_top=True,
weights=None,
classes=num_classes)
elif arch_lower == "resnet101":
model = ResNet101(include_top=True,
weights=None,
classes=num_classes)
elif arch_lower == "resnet152":
model = ResNet152(include_top=True,
weights=None,
classes=num_classes)
elif arch_lower == "mobilenet":
model = MobileNet(include_top=True,
weights=None,
classes=num_classes)
elif arch_lower == "mobilenetv2":
model = MobileNetV2(include_top=True,
weights=None,
classes=num_classes)
elif arch_lower == "densenet121":
model = DenseNet121(include_top=True,
weights=None,
classes=num_classes)
elif arch_lower == "densenet169":
model = DenseNet169(include_top=True,
weights=None,
classes=num_classes)
elif arch_lower == "densenet201":
model = DenseNet201(include_top=True,
weights=None,
classes=num_classes)
else:
raise ModuleNotFoundError("arch \'" + pretrained.get('arch') +
"\' not supported.")
else:
model = Sequential()
# model.add(
# Conv2D(
# 64,
# kernel_size=3,
# activation="relu",
# input_shape=input_shape))
# model.add(MaxPooling2D(pool_size=(2, 2)))
# model.add(Conv2D(64, kernel_size=3, activation="relu"))
# model.add(MaxPooling2D(pool_size=(2, 2)))
# model.add(Flatten())
# model.add(Dense(num_classes, activation="softmax"))
# model.compile(
# optimizer="adam",
# loss=loss_func,
# metrics=['accuracy'])
model.add(
Conv2D(filters=64,
kernel_size=5,
activation="relu",
input_shape=input_shape))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Conv2D(filters=64, kernel_size=3, activation="relu"))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(filters=64, kernel_size=3, activation="relu"))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Flatten())
model.add(Dense(units=256, activation="relu"))
model.add(Dropout(0.25))
model.add(Dense(units=num_classes, activation="softmax"))
if pretrained and 'weights' in pretrained and pretrained.get(
'weights') == 'imagenet':
for layer in base_model.layers:
layer.trainable = False
opt = Adam(learning_rate=LR)
model.compile(optimizer=opt, loss=loss_func, metrics=['accuracy'])
logger("Located image data")
if augmentation:
train_data = ImageDataGenerator(rescale=1. / 255,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True)
test_data = ImageDataGenerator(rescale=1. / 255)
logger('Dataset augmented through zoom, shear, flip, and rescale')
else:
train_data = ImageDataGenerator()
test_data = ImageDataGenerator()
logger("->", "Optimal image size identified: {}".format(input_shape))
X_train = train_data.flow_from_directory(
data_path + training_path,
target_size=input_single,
color_mode=color_mode,
batch_size=(16 if processInfo["train_size"] >= 16 else 1),
class_mode=loss_func[:loss_func.find("_")])
X_test = test_data.flow_from_directory(
data_path + testing_path,
target_size=input_single,
color_mode=color_mode,
batch_size=(16 if processInfo["test_size"] >= 16 else 1),
class_mode=loss_func[:loss_func.find("_")])
if epochs <= 0:
raise BaseException("Number of epochs has to be greater than 0.")
print("\n")
logger('Training image model')
# model.summary()
history = model.fit_generator(
X_train,
steps_per_epoch=X_train.n // X_train.batch_size,
validation_data=X_test,
validation_steps=X_test.n // X_test.batch_size,
epochs=epochs,
verbose=verbose)
if fine_tune:
logger(
'->', 'Training accuracy: {}'.format(
history.history['accuracy'][len(history.history['accuracy']) -
1]))
logger(
'->',
'Validation accuracy: {}'.format(history.history['val_accuracy'][
len(history.history['val_accuracy']) - 1]))
for layer in base_model.layers:
layer.trainable = True
opt = Adam(learning_rate=LR / 10)
model.compile(optimizer=opt, loss=loss_func, metrics=['accuracy'])
print("\n\n")
logger('Training fine tuned model')
fine_tuning_epoch = epochs + 10
history_fine = model.fit_generator(
X_train,
steps_per_epoch=X_train.n // X_train.batch_size,
validation_data=X_test,
validation_steps=X_test.n // X_test.batch_size,
epochs=fine_tuning_epoch,
initial_epoch=history.epoch[-1],
verbose=verbose)
#frozen model acc and loss history
acc = history.history['accuracy']
val_acc = history.history['val_accuracy']
loss = history.history['loss']
val_loss = history.history['val_loss']
#fine tuned model acc and loss history
acc += history_fine.history['accuracy']
val_acc += history_fine.history['val_accuracy']
loss += history_fine.history['loss']
val_loss += history_fine.history['val_loss']
if generate_plots:
plots = generate_fine_tuned_classification_plots(
acc, val_acc, loss, val_loss, epochs)
models = []
losses = []
accuracies = []
model_data = []
model_data.append(model)
models.append(history)
losses.append(
history.history["val_loss"][len(history.history["val_loss"]) - 1])
accuracies.append(
history.history['val_accuracy'][len(history.history['val_accuracy']) -
1])
# final_model = model_data[accuracies.index(max(accuracies))]
# final_hist = models[accuracies.index(max(accuracies))]
if generate_plots and not fine_tune:
plots = generate_classification_plots(models[len(models) - 1])
print("\n")
logger(
'->', 'Final training accuracy: {}'.format(
history.history['accuracy'][len(history.history['accuracy']) - 1]))
logger(
'->',
'Final validation accuracy: {}'.format(history.history['val_accuracy'][
len(history.history['val_accuracy']) - 1]))
# storing values the model dictionary
number_of_examples = len(X_test.filenames)
number_of_generator_calls = math.ceil(number_of_examples /
(1.0 * X_test.batch_size))
test_labels = []
for i in range(0, int(number_of_generator_calls)):
test_labels.extend(np.array(X_test[i][1]))
predIdx = model.predict(X_test)
if output_layer_activation == "sigmoid":
real = [int(x) for x in test_labels]
ans = []
for i in range(len(predIdx)):
ans.append(int(round(predIdx[i][0])))
elif output_layer_activation == "softmax":
real = []
for ans in test_labels:
real.append(ans.argmax())
ans = []
for r in predIdx:
ans.append(r.argmax())
else:
print("NOT THE CASE")
logger("Stored model under 'convolutional_NN' key")
if save_as_tfjs:
tfjs.converters.save_keras_model(model, "tfjsmodel")
logger("Saved tfjs model under 'tfjsmodel' directory")
if save_as_tflite:
converter = tf.lite.TFLiteConverter.from_keras_model(model)
tflite_model = converter.convert()
open("model.tflite", "wb").write(tflite_model)
logger("Saved tflite model as 'model.tflite' ")
clearLog()
K.clear_session()
return {
'id': generate_id(),
'data_type': read_mode,
'data_path': data_path,
'data': {
'train': X_train,
'test': X_test
},
'shape': input_shape,
'res': {
'real': real,
'ans': ans
},
'model': model,
'plots': plots,
'losses': {
'training_loss': history.history['loss'],
'val_loss': history.history['val_loss']
},
'accuracy': {
'training_accuracy': history.history['accuracy'],
'validation_accuracy': history.history['val_accuracy']
},
'num_classes': (2 if num_classes == 1 else num_classes),
'data_sizes': {
'train_size': processInfo['train_size'],
'test_size': processInfo['test_size']
}
}
def classification_ann(instruction,
dataset=None,
text=None,
ca_threshold=None,
preprocess=True,
callback_mode='min',
drop=None,
random_state=49,
test_size=0.2,
epochs=50,
generate_plots=True,
maximizer="val_loss",
save_model=True,
save_path=os.getcwd()):
global currLog
logger("Reading in dataset...")
dataReader = DataReader(dataset)
data = dataReader.data_generator()
if drop is not None:
data.drop(drop, axis=1, inplace=True)
data, y, remove, full_pipeline = initial_preprocesser(
data, instruction, preprocess, ca_threshold, text)
logger("->", "Target Column Found: {}".format(remove))
# Needed to make a custom label encoder due to train test split changes
# Can still be inverse transformed, just a bit of extra work
y = pd.concat([y['train'], y['test']], axis=0)
num_classes = len(np.unique(y))
X_train = data['train']
X_test = data['test']
# ANN needs target one hot encoded for classification
one_hot_encoder = OneHotEncoder()
y = pd.DataFrame(one_hot_encoder.fit_transform(
np.reshape(y.values, (-1, 1))).toarray(),
columns=one_hot_encoder.get_feature_names())
y_train = y.iloc[:len(X_train)]
y_test = y.iloc[len(X_train):]
models = []
losses = []
accuracies = []
model_data = []
logger("Establishing callback function...")
# early stopping callback
es = EarlyStopping(monitor=maximizer, mode='min', verbose=0, patience=5)
i = 0
model = get_keras_model_class(data, i, num_classes)
logger("Training initial model...")
history = model.fit(X_train,
y_train,
epochs=epochs,
validation_data=(X_test, y_test),
callbacks=[es],
verbose=0)
model_data.append(model)
models.append(history)
col_name = [[
"Initial number of layers ", "| Training Loss ", "| Test Loss "
]]
col_width = max(len(word) for row in col_name for word in row) + 2
for row in col_name:
print((" " * 2 * counter) + "| " +
("".join(word.ljust(col_width) for word in row)) + " |")
values = []
values.append(str(len(model.layers)))
values.append(
"| " +
str(history.history['loss'][len(history.history['val_loss']) - 1]))
values.append(
"| " +
str(history.history['val_loss'][len(history.history['val_loss']) - 1]))
datax = []
datax.append(values)
for row in datax:
print((" " * 2 * counter) + "| " +
("".join(word.ljust(col_width) for word in row)) + " |")
#print((" " * 2 * counter)+ tabulate(datax, headers=col_name, tablefmt='orgtbl'))
losses.append(history.history[maximizer][len(history.history[maximizer]) -
1])
# keeps running model and fit functions until the validation loss stops
# decreasing
logger("Testing number of layers...")
col_name = [["Current number of layers", "| Training Loss", "| Test Loss"]]
col_width = max(len(word) for row in col_name for word in row) + 2
for row in col_name:
print((" " * 2 * counter) + "| " +
("".join(word.ljust(col_width) for word in row)) + " |")
datax = []
while (all(x > y for x, y in zip(losses, losses[1:]))):
model = get_keras_model_class(data, i, num_classes)
history = model.fit(X_train,
y_train,
epochs=epochs,
validation_data=(X_test, y_test),
callbacks=[es],
verbose=0)
values = []
datax = []
values.append(str(len(model.layers)))
values.append(
"| " +
str(history.history['loss'][len(history.history['val_loss']) - 1]))
values.append("| " + str(history.history['val_loss'][
len(history.history['val_loss']) - 1]))
datax.append(values)
for row in datax:
print((" " * 2 * counter) + "| " +
("".join(word.ljust(col_width) for word in row)) + " |")
losses.append(
history.history[maximizer][len(history.history[maximizer]) - 1])
accuracies.append(history.history['val_accuracy'][
len(history.history['val_accuracy']) - 1])
i += 1
#print((" " * 2 * counter)+ tabulate(datax, headers=col_name, tablefmt='orgtbl'))
#del values, datax
final_model = model_data[losses.index(min(losses))]
final_hist = models[losses.index(min(losses))]
print("")
logger('->',
"Best number of layers found: " + str(len(final_model.layers)))
logger(
'->', "Training Accuracy: " + str(final_hist.history['accuracy'][
len(final_hist.history['val_accuracy']) - 1]))
logger(
'->', "Test Accuracy: " + str(final_hist.history['val_accuracy'][
len(final_hist.history['val_accuracy']) - 1]))
# genreates appropriate classification plots by feeding all information
plots = generate_classification_plots(models[len(models) - 1], data, y,
model, X_test, y_test)
if save_model:
save(final_model, save_model)
print("")
logger("Stored model under 'classification_ANN' key")
# stores the values and plots into the object dictionary
return {
'id': generate_id(),
"model": final_model,
'num_classes': num_classes,
"plots": plots,
"target": remove,
"preprocesser": full_pipeline,
"interpreter": one_hot_encoder,
'losses': {
'training_loss': final_hist.history['loss'],
'val_loss': final_hist.history['val_loss']
},
'accuracy': {
'training_accuracy': final_hist.history['accuracy'],
'validation_accuracy': final_hist.history['val_accuracy']
}
}
