Loading deepchem/hyperparameters/__init__.py +2 −0 Original line number Diff line number Diff line Loading @@ -22,6 +22,8 @@ class HyperparamOpt(object): assert verbosity in [None, "low", "high"] self.verbosity = verbosity # TODO(rbharath): This function is complicated and monolithic. Is there a nice # way to refactor this? def hyperparam_search(self, params_dict, train_dataset, valid_dataset, output_transformers, metric, use_max=True, logdir=None): Loading deepchem/hyperparameters/tests/test_hyperparam_opt.py +34 −1 Original line number Diff line number Diff line Loading @@ -20,7 +20,8 @@ from deepchem.featurizers.fingerprints import CircularFingerprint from deepchem.transformers import NormalizationTransformer from deepchem import metrics from deepchem.metrics import Metric from sklearn.ensemble import RandomForestRegressor from deepchem.models.multitask import SingletaskToMultitask from sklearn.linear_model import LogisticRegression def rf_model_builder(task_types, params_dict, logdir=None, verbosity=None): """Builds random forests given hyperparameters. Loading Loading @@ -63,6 +64,38 @@ class TestHyperparamOptAPI(TestAPI): valid_dataset, output_transformers, task_types, metric) def test_singletask_to_multitask_sklearn_hyperparam_opt(self): """Test of hyperparam_opt with singletask_to_multitask.""" splittype = "scaffold" compound_featurizers = [CircularFingerprint(size=1024)] complex_featurizers = [] output_transformer_classes = [] input_transformer_classes = [] tasks = ["task0", "task1", "task2", "task3", "task4", "task5", "task6", "task7", "task8", "task9", "task10", "task11", "task12", "task13", "task14", "task15", "task16"] task_types = {task: "classification" for task in tasks} input_file = "multitask_example.csv" train_dataset, valid_dataset, _, output_transformers, = \ self._featurize_train_test_split( splittype, compound_featurizers, complex_featurizers, input_transformer_classes, output_transformer_classes, input_file, task_types.keys()) params_dict = { "batch_size": [32], "data_shape": [train_dataset.get_data_shape()], } classification_metric = Metric(metrics.matthews_corrcoef, np.mean) def model_builder(task_types, model_params, verbosity=None): return SklearnModel(task_types, model_params, model_instance=LogisticRegression()) def multitask_model_builder(task_types, params_dict, logdir=None, verbosity=None): return SingletaskToMultitask(task_types, params_dict, model_builder) self._hyperparam_opt(multitask_model_builder, params_dict, train_dataset, valid_dataset, output_transformers, task_types, classification_metric) def test_multitask_keras_mlp_ECFP_classification_hyperparam_opt(self): """Straightforward test of Keras multitask deepchem classification API.""" from deepchem.models.keras_models.fcnet import MultiTaskDNN Loading deepchem/models/__init__.py +0 −1 Original line number Diff line number Diff line Loading @@ -27,7 +27,6 @@ class Model(object): """ Abstract base class for different ML models. """ non_sklearn_models = ["SingleTaskDNN", "MultiTaskDNN", "DockingDNN"] def __init__(self, task_types, model_params, fit_transformers=None, model_instance=None, initialize_raw_model=True, verbosity=None, **kwargs): Loading deepchem/models/multitask.py 0 → 100644 +47 −0 Original line number Diff line number Diff line """ Convenience class that lets singletask models fit on multitask data. """ from __future__ import print_function from __future__ import division from __future__ import unicode_literals import numpy as np from deepchem.models import Model class SingletaskToMultitask(Model): """ Convenience class to let singletask models be fit on multitask data. Warning: This current implementation is only functional for sklearn models. """ def __init__(self, task_types, model_params, model_builder, verbosity=None): self.task_types = task_types self.model_params = model_params self.models = {} self.fit_transformers = False for task, task_type in self.task_types.iteritems(): self.models[task] = model_builder(task_types, model_params, verbosity=verbosity) def fit(self, dataset): """ Updates all singletask models with new information. Warning: This current implementation is only functional for sklearn models. """ X, y, _, _ = dataset.to_numpy() for ind, task in enumerate(self.task_types.keys()): y_task = y[:, ind] self.models[task].raw_model.fit(X, y_task) def predict_on_batch(self, X): """ Concatenates results from all singletask models. """ N_tasks = len(self.task_types.keys()) N_samples = X.shape[0] y_pred = np.zeros((N_samples, N_tasks)) for ind, task in enumerate(self.task_types.keys()): y_pred[:, ind] = self.models[task].predict_on_batch(X) return y_pred deepchem/models/test/test_singletask_to_multitask.py 0 → 100644 +50 −0 Original line number Diff line number Diff line """ Testing singletask-to-multitask. """ from __future__ import print_function from __future__ import division from __future__ import unicode_literals __author__ = "Bharath Ramsundar" __copyright__ = "Copyright 2016, Stanford University" __license__ = "LGPL" from deepchem.models.test import TestAPI from deepchem import metrics from deepchem.metrics import Metric from deepchem.featurizers.fingerprints import CircularFingerprint from deepchem.models.multitask import SingletaskToMultitask from deepchem.models.sklearn_models import SklearnModel from sklearn.linear_model import LogisticRegression class TestSingletasktoMultitaskAPI(TestAPI): """ Test top-level API for singletask_to_multitask ML models. """ def test_singletask_to_multitask_classification(self): splittype = "scaffold" compound_featurizers = [CircularFingerprint(size=1024)] complex_featurizers = [] output_transformer_classes = [] input_transformer_classes = [] tasks = ["task0", "task1", "task2", "task3", "task4", "task5", "task6", "task7", "task8", "task9", "task10", "task11", "task12", "task13", "task14", "task15", "task16"] task_types = {task: "classification" for task in tasks} input_file = "multitask_example.csv" train_dataset, test_dataset, _, output_transformers, = \ self._featurize_train_test_split( splittype, compound_featurizers, complex_featurizers, input_transformer_classes, output_transformer_classes, input_file, task_types.keys()) params_dict = { "batch_size": 32, "data_shape": train_dataset.get_data_shape() } classification_metrics = [Metric(metrics.roc_auc_score)] def model_builder(task_types, model_params, verbosity=None): return SklearnModel(task_types, model_params, model_instance=LogisticRegression()) multitask_model = SingletaskToMultitask(task_types, params_dict, model_builder) self._create_model(train_dataset, test_dataset, multitask_model, output_transformers, classification_metrics) Loading
deepchem/hyperparameters/__init__.py +2 −0 Original line number Diff line number Diff line Loading @@ -22,6 +22,8 @@ class HyperparamOpt(object): assert verbosity in [None, "low", "high"] self.verbosity = verbosity # TODO(rbharath): This function is complicated and monolithic. Is there a nice # way to refactor this? def hyperparam_search(self, params_dict, train_dataset, valid_dataset, output_transformers, metric, use_max=True, logdir=None): Loading
deepchem/hyperparameters/tests/test_hyperparam_opt.py +34 −1 Original line number Diff line number Diff line Loading @@ -20,7 +20,8 @@ from deepchem.featurizers.fingerprints import CircularFingerprint from deepchem.transformers import NormalizationTransformer from deepchem import metrics from deepchem.metrics import Metric from sklearn.ensemble import RandomForestRegressor from deepchem.models.multitask import SingletaskToMultitask from sklearn.linear_model import LogisticRegression def rf_model_builder(task_types, params_dict, logdir=None, verbosity=None): """Builds random forests given hyperparameters. Loading Loading @@ -63,6 +64,38 @@ class TestHyperparamOptAPI(TestAPI): valid_dataset, output_transformers, task_types, metric) def test_singletask_to_multitask_sklearn_hyperparam_opt(self): """Test of hyperparam_opt with singletask_to_multitask.""" splittype = "scaffold" compound_featurizers = [CircularFingerprint(size=1024)] complex_featurizers = [] output_transformer_classes = [] input_transformer_classes = [] tasks = ["task0", "task1", "task2", "task3", "task4", "task5", "task6", "task7", "task8", "task9", "task10", "task11", "task12", "task13", "task14", "task15", "task16"] task_types = {task: "classification" for task in tasks} input_file = "multitask_example.csv" train_dataset, valid_dataset, _, output_transformers, = \ self._featurize_train_test_split( splittype, compound_featurizers, complex_featurizers, input_transformer_classes, output_transformer_classes, input_file, task_types.keys()) params_dict = { "batch_size": [32], "data_shape": [train_dataset.get_data_shape()], } classification_metric = Metric(metrics.matthews_corrcoef, np.mean) def model_builder(task_types, model_params, verbosity=None): return SklearnModel(task_types, model_params, model_instance=LogisticRegression()) def multitask_model_builder(task_types, params_dict, logdir=None, verbosity=None): return SingletaskToMultitask(task_types, params_dict, model_builder) self._hyperparam_opt(multitask_model_builder, params_dict, train_dataset, valid_dataset, output_transformers, task_types, classification_metric) def test_multitask_keras_mlp_ECFP_classification_hyperparam_opt(self): """Straightforward test of Keras multitask deepchem classification API.""" from deepchem.models.keras_models.fcnet import MultiTaskDNN Loading
deepchem/models/__init__.py +0 −1 Original line number Diff line number Diff line Loading @@ -27,7 +27,6 @@ class Model(object): """ Abstract base class for different ML models. """ non_sklearn_models = ["SingleTaskDNN", "MultiTaskDNN", "DockingDNN"] def __init__(self, task_types, model_params, fit_transformers=None, model_instance=None, initialize_raw_model=True, verbosity=None, **kwargs): Loading
deepchem/models/multitask.py 0 → 100644 +47 −0 Original line number Diff line number Diff line """ Convenience class that lets singletask models fit on multitask data. """ from __future__ import print_function from __future__ import division from __future__ import unicode_literals import numpy as np from deepchem.models import Model class SingletaskToMultitask(Model): """ Convenience class to let singletask models be fit on multitask data. Warning: This current implementation is only functional for sklearn models. """ def __init__(self, task_types, model_params, model_builder, verbosity=None): self.task_types = task_types self.model_params = model_params self.models = {} self.fit_transformers = False for task, task_type in self.task_types.iteritems(): self.models[task] = model_builder(task_types, model_params, verbosity=verbosity) def fit(self, dataset): """ Updates all singletask models with new information. Warning: This current implementation is only functional for sklearn models. """ X, y, _, _ = dataset.to_numpy() for ind, task in enumerate(self.task_types.keys()): y_task = y[:, ind] self.models[task].raw_model.fit(X, y_task) def predict_on_batch(self, X): """ Concatenates results from all singletask models. """ N_tasks = len(self.task_types.keys()) N_samples = X.shape[0] y_pred = np.zeros((N_samples, N_tasks)) for ind, task in enumerate(self.task_types.keys()): y_pred[:, ind] = self.models[task].predict_on_batch(X) return y_pred
deepchem/models/test/test_singletask_to_multitask.py 0 → 100644 +50 −0 Original line number Diff line number Diff line """ Testing singletask-to-multitask. """ from __future__ import print_function from __future__ import division from __future__ import unicode_literals __author__ = "Bharath Ramsundar" __copyright__ = "Copyright 2016, Stanford University" __license__ = "LGPL" from deepchem.models.test import TestAPI from deepchem import metrics from deepchem.metrics import Metric from deepchem.featurizers.fingerprints import CircularFingerprint from deepchem.models.multitask import SingletaskToMultitask from deepchem.models.sklearn_models import SklearnModel from sklearn.linear_model import LogisticRegression class TestSingletasktoMultitaskAPI(TestAPI): """ Test top-level API for singletask_to_multitask ML models. """ def test_singletask_to_multitask_classification(self): splittype = "scaffold" compound_featurizers = [CircularFingerprint(size=1024)] complex_featurizers = [] output_transformer_classes = [] input_transformer_classes = [] tasks = ["task0", "task1", "task2", "task3", "task4", "task5", "task6", "task7", "task8", "task9", "task10", "task11", "task12", "task13", "task14", "task15", "task16"] task_types = {task: "classification" for task in tasks} input_file = "multitask_example.csv" train_dataset, test_dataset, _, output_transformers, = \ self._featurize_train_test_split( splittype, compound_featurizers, complex_featurizers, input_transformer_classes, output_transformer_classes, input_file, task_types.keys()) params_dict = { "batch_size": 32, "data_shape": train_dataset.get_data_shape() } classification_metrics = [Metric(metrics.roc_auc_score)] def model_builder(task_types, model_params, verbosity=None): return SklearnModel(task_types, model_params, model_instance=LogisticRegression()) multitask_model = SingletaskToMultitask(task_types, params_dict, model_builder) self._create_model(train_dataset, test_dataset, multitask_model, output_transformers, classification_metrics)
