def test_io_api(tmp_path):
num_instances = 100
(image_x, train_y), (test_x, test_y) = mnist.load_data()
(text_x, train_y), (test_x,
test_y) = utils.imdb_raw(num_instances=num_instances)
image_x = image_x[:num_instances]
text_x = text_x[:num_instances]
structured_data_x = utils.generate_structured_data(
num_instances=num_instances)
classification_y = utils.generate_one_hot_labels(
num_instances=num_instances, num_classes=3)
regression_y = utils.generate_data(num_instances=num_instances,
shape=(1, ))
# Build model and train.
automodel = ak.AutoModel(
inputs=[ak.ImageInput(),
ak.TextInput(),
ak.StructuredDataInput()],
outputs=[
ak.RegressionHead(metrics=['mae']),
ak.ClassificationHead(loss='categorical_crossentropy',
metrics=['accuracy'])
],
directory=tmp_path,
max_trials=2,
seed=utils.SEED)
automodel.fit([image_x, text_x, structured_data_x],
[regression_y, classification_y],
epochs=1,
validation_split=0.2)
def applyAutoKeras(X_train, y_train, X_test, y_test, SavePath,
max_trials=100, epochs=300, useSavedModels = True):
if not useSavedModels or not os.path.isdir(SavePath+"/keras_auto_model/best_model/"):
input_node = ak.StructuredDataInput()
output_node = ak.DenseBlock()(input_node)
#output_node = ak.ConvBlock()(output_node)
output_node = ak.RegressionHead()(output_node)
AKRegressor = ak.AutoModel(
inputs=input_node,
outputs=output_node,
max_trials=max_trials,
overwrite=True,
tuner="bayesian",
project_name=SavePath+"/keras_auto_model"
)
print(" X_train shape: {0}\n y_train shape: {1}\n X_test shape: {2}\n y_test shape: {3}".format(X_train.shape, y_train.shape, X_test.shape, y_test.shape))
AKRegressor.fit(x=X_train, y=y_train[:,0],epochs=epochs,verbose=1, batch_size=int(X_train.shape[0]/10), shuffle=False, use_multiprocessing=True)
AKRegressor.export_model()
else:
AKRegressor = tf.keras.models.load_model(SavePath+"/keras_auto_model/best_model/")
y_hat = AKRegressor.predict(X_test)
print("AUTOKERAS - Score: ")
print("MAE: %.4f" % mean_absolute_error(y_test[:,0], y_hat))
return y_hat
def test_io_api(tmp_path):
num_instances = 20
image_x = utils.generate_data(num_instances=num_instances, shape=(28, 28))
text_x = utils.generate_text_data(num_instances=num_instances)
image_x = image_x[:num_instances]
structured_data_x = (pd.read_csv(utils.TRAIN_CSV_PATH).to_numpy().astype(
np.unicode)[:num_instances])
classification_y = utils.generate_one_hot_labels(
num_instances=num_instances, num_classes=3)
regression_y = utils.generate_data(num_instances=num_instances,
shape=(1, ))
# Build model and train.
automodel = ak.AutoModel(
inputs=[ak.ImageInput(),
ak.TextInput(),
ak.StructuredDataInput()],
outputs=[
ak.RegressionHead(metrics=["mae"]),
ak.ClassificationHead(loss="categorical_crossentropy",
metrics=["accuracy"]),
],
directory=tmp_path,
max_trials=2,
tuner=ak.RandomSearch,
seed=utils.SEED,
)
automodel.fit(
[image_x, text_x, structured_data_x],
[regression_y, classification_y],
epochs=1,
validation_split=0.2,
batch_size=4,
)
def test_feature_engineering():
dataset = common.generate_structured_data(dtype='dataset')
feature = preprocessor_module.FeatureEngineering()
feature.input_node = ak.StructuredDataInput(
column_names=common.COLUMN_NAMES_FROM_NUMPY,
column_types=common.COLUMN_TYPES_FROM_NUMPY)
new_dataset = run_preprocessor(feature, dataset,
common.generate_data(dtype='dataset'),
tf.float32)
assert isinstance(new_dataset, tf.data.Dataset)
def train():
house_dataset = fetch_california_housing()
df = pd.DataFrame(np.concatenate(
(house_dataset.data, house_dataset.target.reshape(-1, 1)), axis=1),
columns=house_dataset.feature_names + ['Price'])
train_size = int(df.shape[0] * 0.9)
df[:train_size].to_csv('train.csv', index=False)
df[train_size:].to_csv('eval.csv', index=False)
train_file_path = 'train.csv'
test_file_path = 'eval.csv'
# x_train as pandas.DataFrame, y_train as pandas.Series
x_train = pd.read_csv(train_file_path)
print(type(x_train)) # pandas.DataFrame
y_train = x_train.pop('Price')
print(type(y_train)) # pandas.Series
# You can also use pandas.DataFrame for y_train.
y_train = pd.DataFrame(y_train)
print(type(y_train)) # pandas.DataFrame
# You can also use numpy.ndarray for x_train and y_train.
x_train = x_train.to_numpy().astype(np.unicode)
y_train = y_train.to_numpy()
print(type(x_train)) # numpy.ndarray
print(type(y_train)) # numpy.ndarray
# Preparing testing data.
x_test = pd.read_csv(test_file_path)
y_test = x_test.pop('Price')
# Initialize the structured data regressor.
input_node = ak.StructuredDataInput()
output_node = ak.StructuredDataBlock(
categorical_encoding=True)(input_node)
output_node = ak.RegressionHead()(output_node)
reg = ak.AutoModel(inputs=input_node,
outputs=output_node,
overwrite=True,
max_trials=3)
# Feed the structured data regressor with training data.
reg.fit(
x_train,
y_train,
# Split the training data and use the last 15% as validation data.
validation_split=0.15,
epochs=10)
# Predict with the best model.
predicted_y = reg.predict(test_file_path)
# Evaluate the best model with testing data.
print(reg.evaluate(test_file_path, 'Price'))
def test_functional_api(tmp_dir):
# Prepare the data.
num_instances = 20
(image_x, train_y), (test_x, test_y) = mnist.load_data()
(text_x, train_y), (test_x, test_y) = common.imdb_raw()
(structured_data_x, train_y), (test_x, test_y) = common.dataframe_numpy()
image_x = image_x[:num_instances]
text_x = text_x[:num_instances]
structured_data_x = structured_data_x[:num_instances]
classification_y = common.generate_one_hot_labels(
num_instances=num_instances, num_classes=3)
regression_y = common.generate_data(num_instances=num_instances,
shape=(1, ))
# Build model and train.
image_input = ak.ImageInput()
output = ak.Normalization()(image_input)
output = ak.ImageAugmentation()(output)
outputs1 = ak.ResNetBlock(version='next')(image_input)
outputs2 = ak.XceptionBlock()(image_input)
image_output = ak.Merge()((outputs1, outputs2))
structured_data_input = ak.StructuredDataInput(
column_names=common.COLUMN_NAMES_FROM_CSV,
column_types=common.COLUMN_TYPES_FROM_CSV)
structured_data_output = ak.FeatureEngineering()(structured_data_input)
structured_data_output = ak.DenseBlock()(structured_data_output)
text_input = ak.TextInput()
outputs1 = ak.TextToIntSequence()(text_input)
outputs1 = ak.EmbeddingBlock()(outputs1)
outputs1 = ak.ConvBlock(separable=True)(outputs1)
outputs1 = ak.SpatialReduction()(outputs1)
outputs2 = ak.TextToNgramVector()(text_input)
outputs2 = ak.DenseBlock()(outputs2)
text_output = ak.Merge()((outputs1, outputs2))
merged_outputs = ak.Merge()(
(structured_data_output, image_output, text_output))
regression_outputs = ak.RegressionHead()(merged_outputs)
classification_outputs = ak.ClassificationHead()(merged_outputs)
automodel = ak.GraphAutoModel(
inputs=[image_input, text_input, structured_data_input],
directory=tmp_dir,
outputs=[regression_outputs, classification_outputs],
max_trials=2,
seed=common.SEED)
automodel.fit((image_x, text_x, structured_data_x),
(regression_y, classification_y),
validation_split=0.2,
epochs=2)
def test_functional_api(tmp_path):
# Prepare the data.
num_instances = 80
(image_x, train_y), (test_x, test_y) = mnist.load_data()
(text_x, train_y), (test_x, test_y) = utils.imdb_raw()
(structured_data_x, train_y), (test_x, test_y) = utils.dataframe_numpy()
image_x = image_x[:num_instances]
text_x = text_x[:num_instances]
structured_data_x = structured_data_x[:num_instances]
classification_y = utils.generate_one_hot_labels(
num_instances=num_instances, num_classes=3)
regression_y = utils.generate_data(num_instances=num_instances,
shape=(1, ))
# Build model and train.
image_input = ak.ImageInput()
output = ak.Normalization()(image_input)
output = ak.ImageAugmentation()(output)
outputs1 = ak.ResNetBlock(version='next')(output)
outputs2 = ak.XceptionBlock()(output)
image_output = ak.Merge()((outputs1, outputs2))
structured_data_input = ak.StructuredDataInput()
structured_data_output = ak.CategoricalToNumerical()(structured_data_input)
structured_data_output = ak.DenseBlock()(structured_data_output)
text_input = ak.TextInput()
outputs1 = ak.TextToIntSequence()(text_input)
outputs1 = ak.Embedding()(outputs1)
outputs1 = ak.ConvBlock(separable=True)(outputs1)
outputs1 = ak.SpatialReduction()(outputs1)
outputs2 = ak.TextToNgramVector()(text_input)
outputs2 = ak.DenseBlock()(outputs2)
text_output = ak.Merge()((outputs1, outputs2))
merged_outputs = ak.Merge()(
(structured_data_output, image_output, text_output))
regression_outputs = ak.RegressionHead()(merged_outputs)
classification_outputs = ak.ClassificationHead()(merged_outputs)
automodel = ak.AutoModel(
inputs=[image_input, text_input, structured_data_input],
directory=tmp_path,
outputs=[regression_outputs, classification_outputs],
max_trials=2,
tuner=ak.Hyperband,
seed=utils.SEED)
automodel.fit((image_x, text_x, structured_data_x),
(regression_y, classification_y),
validation_split=0.2,
epochs=1)
def test_structured_data_block():
block = wrapper.StructuredDataBlock()
block.num_heads = 1
block.column_names = ['0', '1']
block.column_types = {
'0': adapters.CATEGORICAL,
'1': adapters.CATEGORICAL,
}
hp = kerastuner.HyperParameters()
output = block.build(hp, ak.StructuredDataInput(shape=(2, )).build())
assert isinstance(output, tf.Tensor)
def train():
TRAIN_DATA_URL = "https://storage.googleapis.com/tf-datasets/titanic/train.csv"
TEST_DATA_URL = "https://storage.googleapis.com/tf-datasets/titanic/eval.csv"
train_file_path = tf.keras.utils.get_file("train.csv", TRAIN_DATA_URL)
test_file_path = tf.keras.utils.get_file("eval.csv", TEST_DATA_URL)
# x_train as pandas.DataFrame, y_train as pandas.Series
x_train = pd.read_csv(train_file_path)
print(type(x_train)) # pandas.DataFrame
y_train = x_train.pop('survived')
print(type(y_train)) # pandas.Series
# You can also use pandas.DataFrame for y_train.
y_train = pd.DataFrame(y_train)
print(type(y_train)) # pandas.DataFrame
# You can also use numpy.ndarray for x_train and y_train.
x_train = x_train.to_numpy().astype(np.unicode)
y_train = y_train.to_numpy()
print(type(x_train)) # numpy.ndarray
print(type(y_train)) # numpy.ndarray
# Preparing testing data.
x_test = pd.read_csv(test_file_path)
y_test = x_test.pop('survived')
# Initialize the structured data classifier.
input_node = ak.StructuredDataInput()
output_node = ak.StructuredDataBlock(
categorical_encoding=True)(input_node)
output_node = ak.ClassificationHead()(output_node)
clf = ak.AutoModel(inputs=input_node,
outputs=output_node,
overwrite=True,
max_trials=3)
# Feed the structured data classifier with training data.
clf.fit(
x_train,
y_train,
# Split the training data and use the last 15% as validation data.
validation_split=0.15,
epochs=10)
# Predict with the best model.
predicted_y = clf.predict(test_file_path)
# Evaluate the best model with testing data.
print(clf.evaluate(test_file_path, 'survived'))
def test_multi_model():
context = an.AutoMLPipeline(
an.MultiModel(
inputs=[ak.ImageInput(), ak.StructuredDataInput()],
outputs=[
ak.RegressionHead(metrics=["mae"]),
ak.ClassificationHead(loss="categorical_crossentropy",
metrics=["accuracy"]),
],
overwrite=True,
max_trials=2,
))
context.run_automl()
assert context.return_automl["model"] != None
def test_text_and_structured_data(tmp_path):
# Prepare the data.
num_instances = 80
(x_text, y_train), (x_test, y_test) = utils.imdb_raw()
x_structured_data = pd.read_csv(utils.TRAIN_CSV_PATH)
x_text = x_text[:num_instances]
x_structured_data = x_structured_data[:num_instances]
y_classification = utils.generate_one_hot_labels(
num_instances=num_instances, num_classes=3)
y_regression = utils.generate_data(num_instances=num_instances,
shape=(1, ))
# Build model and train.
structured_data_input = ak.StructuredDataInput()
structured_data_output = ak.CategoricalToNumerical()(structured_data_input)
structured_data_output = ak.DenseBlock()(structured_data_output)
text_input = ak.TextInput()
outputs1 = ak.TextToIntSequence()(text_input)
outputs1 = ak.Embedding()(outputs1)
outputs1 = ak.ConvBlock(separable=True)(outputs1)
outputs1 = ak.SpatialReduction()(outputs1)
outputs2 = ak.TextToNgramVector()(text_input)
outputs2 = ak.DenseBlock()(outputs2)
text_output = ak.Merge()((outputs1, outputs2))
merged_outputs = ak.Merge()((structured_data_output, text_output))
regression_outputs = ak.RegressionHead()(merged_outputs)
classification_outputs = ak.ClassificationHead()(merged_outputs)
automodel = ak.AutoModel(
inputs=[text_input, structured_data_input],
directory=tmp_path,
outputs=[regression_outputs, classification_outputs],
max_trials=2,
tuner=ak.Hyperband,
seed=utils.SEED,
)
automodel.fit(
(x_text, x_structured_data),
(y_regression, y_classification),
validation_split=0.2,
epochs=1,
)
def test_feature_engineering_fix_keyerror():
data = structured_data(100)
dataset = tf.data.Dataset.from_tensor_slices(data)
feature = preprocessor.FeatureEngineering()
feature.input_node = ak.StructuredDataInput(
column_names=COLUMN_NAMES_FROM_NUMPY,
column_types=COLUMN_TYPES_FROM_NUMPY)
feature.set_hp(kerastuner.HyperParameters())
for x in dataset:
feature.update(x)
feature.finalize()
feature.set_config(feature.get_config())
for a in dataset:
feature.transform(a)
def map_func(x):
return tf.py_function(feature.transform, inp=[x], Tout=(tf.float64, ))
new_dataset = dataset.map(map_func)
assert isinstance(new_dataset, tf.data.Dataset)
def test_structured_data_input(tmp_dir):
num_data = 500
data = common.structured_data(num_data)
x_train = data
y = np.random.randint(0, 3, num_data)
y_train = y
input_node = ak.StructuredDataInput(
column_names=common.COLUMN_NAMES_FROM_NUMPY,
column_types=common.COLUMN_TYPES_FROM_NUMPY)
output_node = input_node
output_node = ak.StructuredDataBlock()(output_node)
output_node = ak.ClassificationHead(loss='categorical_crossentropy',
metrics=['accuracy'])(output_node)
auto_model = ak.GraphAutoModel(input_node,
output_node,
directory=tmp_dir,
max_trials=1)
auto_model.fit(x_train,
y_train,
epochs=1,
validation_data=(x_train, y_train))
auto_model.predict(x_train)
# Generate regression targets.
regression_target = np.random.rand(num_instances, 1).astype(np.float32)
# Generate classification labels of five classes.
classification_target = np.random.randint(5, size=num_instances)
"""
## Build and Train the Model
Then we initialize the multi-modal and multi-task model with
[AutoModel](/auto_model/#automodel-class).
Since this is just a demo, we use small amount of `max_trials` and `epochs`.
"""
import autokeras as ak
# Initialize the multi with multiple inputs and outputs.
model = ak.AutoModel(inputs=[ak.ImageInput(),
ak.StructuredDataInput()],
outputs=[
ak.RegressionHead(metrics=['mae']),
ak.ClassificationHead(loss='categorical_crossentropy',
metrics=['accuracy'])
],
overwrite=True,
max_trials=2)
# Fit the model with prepared data.
model.fit([image_data, structured_data],
[regression_target, classification_target],
epochs=3)
"""
## Validation Data
By default, AutoKeras use the last 20% of training data as validation data.
As shown in the example below, you can use `validation_split` to specify the percentage.
x_test = x_test[:data_slice]
y_test = y_test[:data_slice]
x_image = x_train.reshape(x_train.shape + (1,))
x_test = x_test.reshape(x_test.shape + (1,))
x_structured = np.random.rand(x_train.shape[0], 100)
y_regression = np.random.rand(x_train.shape[0], 1)
y_classification = y_classification.reshape(-1, 1)
# Build model and train.
inputs = ak.ImageInput(shape=(28, 28, 1))
outputs1 = ak.ResNetBlock(version='next')(inputs)
outputs2 = ak.XceptionBlock()(inputs)
image_outputs = ak.Merge()((outputs1, outputs2))
structured_inputs = ak.StructuredDataInput()
structured_outputs = ak.DenseBlock()(structured_inputs)
merged_outputs = ak.Merge()((structured_outputs, image_outputs))
classification_outputs = ak.ClassificationHead()(merged_outputs)
regression_outputs = ak.RegressionHead()(merged_outputs)
automodel = ak.GraphAutoModel(inputs=[inputs, structured_inputs],
outputs=[regression_outputs,
classification_outputs])
automodel.fit((x_image, x_structured),
(y_regression, y_classification),
# trials=100,
validation_split=0.2,
epochs=200,
callbacks=[tf.keras.callbacks.EarlyStopping()])
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(x,
y,
test_size=0.33,
random_state=42)
print(X_train.shape)
print(y_train.shape)
print(y_test.shape)
print(y_val.shape)
print(X_train[1])
print(y_train[1])
# In[27]:
id_input = ak.StructuredDataInput()
id_den = ak.CategoricalToNumerical()(id_input)
id_den = ak.Embedding()(id_den)
x_input = ak.Input()
layer = ak.DenseBlock()(x_input)
mer = ak.Merge()([id_den, layer])
output_node = ak.RegressionHead(metrics=['mae'])(mer)
# In[28]:
# auto_model = ak.AutoModel( inputs= x_input,
# #project_name="categorical_model",
# outputs = output_node,
# objective="loss",
validation_data=(x_val, y_val),
epochs=10,
)
"""
## Customized Search Space
For advanced users, you may customize your search space by using
[AutoModel](/auto_model/#automodel-class) instead of
[StructuredDataRegressor](/structured_data_regressor). You can configure the
[StructuredDataBlock](/block/#structureddatablock-class) for some high-level
configurations, e.g., `categorical_encoding` for whether to use the
[CategoricalToNumerical](/block/#categoricaltonumerical-class). You can also do
not specify these arguments, which would leave the different choices to be
tuned automatically. See the following example for detail.
"""
input_node = ak.StructuredDataInput()
output_node = ak.StructuredDataBlock(categorical_encoding=True)(input_node)
output_node = ak.RegressionHead()(output_node)
reg = ak.AutoModel(inputs=input_node,
outputs=output_node,
overwrite=True,
max_trials=3)
reg.fit(x_train, y_train, epochs=10)
"""
The usage of [AutoModel](/auto_model/#automodel-class) is similar to the
[functional API](https://www.tensorflow.org/guide/keras/functional) of Keras.
Basically, you are building a graph, whose edges are blocks and the nodes are
intermediate outputs of blocks. To add an edge from `input_node` to
`output_node` with `output_node = ak.[some_block]([block_args])(input_node)`.
You can even also use more fine grained blocks to customize the search space
# Generate regression targets.
regression_target = np.random.rand(num_instances, 1).astype(np.float32)
# Generate classification labels of five classes.
classification_target = np.random.randint(5, size=num_instances)
"""
## Build and Train the Model
Then we initialize the multi-modal and multi-task model with
[AutoModel](/auto_model/#automodel-class).
"""
import autokeras as ak
# Initialize the multi with multiple inputs and outputs.
model = ak.AutoModel(
inputs=[ak.ImageInput(), ak.StructuredDataInput()],
outputs=[
ak.RegressionHead(metrics=['mae']),
ak.ClassificationHead(loss='categorical_crossentropy', metrics=['accuracy'])
],
max_trials=10)
# Fit the model with prepared data.
model.fit(
[image_data, structured_data],
[regression_target, classification_target],
epochs=10)
"""
## Validation Data
By default, AutoKeras use the last 20% of training data as validation data.
As shown in the example below, you can use `validation_split` to specify the percentage.
(x_train, y_classification), (x_test, y_test) = mnist.load_data()
data_slice = 20
x_train = x_train[:data_slice]
print(x_train.dtype)
y_classification = y_classification[:data_slice]
x_test = x_test[:data_slice]
y_test = y_test[:data_slice]
x_train = x_train.astype(np.float64)
x_test = x_test.astype(np.float64)
print(x_train.dtype)
# x_image = np.reshape(x_train, (200, 28, 28, 1))
# x_test = np.reshape(x_test, (200, 28, 28, 1))
x_image = x_train.reshape(x_train.shape + (1,))
x_test = x_test.reshape(x_test.shape + (1,))
'''x_structured = np.random.rand(x_train.shape[0], 100)
y_regression = np.random.rand(x_train.shape[0], 1)'''
x_structured = np.random.rand(x_train.shape[0], 100)
y_regression = np.random.rand(x_train.shape[0], 1)
y_classification = y_classification.reshape(-1, 1)
# y_classification = np.reshape(y_classification, (-1, 1))
# Build model and train.
automodel = ak.AutoModel(
inputs=[ak.ImageInput(),
ak.StructuredDataInput()],
outputs=[ak.RegressionHead(metrics=['mae']),
ak.ClassificationHead(loss='categorical_crossentropy',
metrics=['accuracy'])])
automodel.fit([x_image, x_structured],
[y_regression, y_classification],
validation_split=0.2)
