new FCN module, model classes can now use external parameters if provided

mlair/model_modules/fully_connected_networks.py

+__author__ = "Lukas Leufen"
+__date__ = '2021-02-'
+
+from mlair.model_modules import AbstractModelClass
+from mlair.helpers import select_from_dict
+
+import keras
+
+
+class FCN_64_32_16(AbstractModelClass):
+    """
+    A customised model 4 Dense layers (64, 32, 16, window_lead_time), where the last layer is the output layer depending
+    on the window_lead_time parameter.
+    """
+
+    def __init__(self, input_shape: list, output_shape: list):
+        """
+        Sets model and loss depending on the given arguments.
+
+        :param input_shape: list of input shapes (expect len=1 with shape=(window_hist, station, variables))
+        :param output_shape: list of output shapes (expect len=1 with shape=(window_forecast))
+        """
+
+        assert len(input_shape) == 1
+        assert len(output_shape) == 1
+        super().__init__(input_shape[0], output_shape[0])
+
+        # settings
+        self.dropout_rate = 0.1
+        self.regularizer = keras.regularizers.l2(0.1)
+        self.activation = keras.layers.PReLU
+
+        # apply to model
+        self.set_model()
+        self.set_compile_options()
+        self.set_custom_objects(loss=self.compile_options['loss'])
+
+    def set_model(self):
+        """
+        Build the model.
+        """
+        x_input = keras.layers.Input(shape=self._input_shape)
+        x_in = keras.layers.Flatten(name='{}'.format("major"))(x_input)
+        x_in = keras.layers.Dense(64, name='{}_Dense_64'.format("major"))(x_in)
+        x_in = self.activation()(x_in)
+        x_in = keras.layers.Dense(32, name='{}_Dense_32'.format("major"))(x_in)
+        x_in = self.activation()(x_in)
+        x_in = keras.layers.Dense(16, name='{}_Dense_16'.format("major"))(x_in)
+        x_in = self.activation()(x_in)
+        x_in = keras.layers.Dense(self._output_shape, name='{}_Dense'.format("major"))(x_in)
+        out_main = self.activation()(x_in)
+        self.model = keras.Model(inputs=x_input, outputs=[out_main])
+
+    def set_compile_options(self):
+        self.initial_lr = 1e-2
+        self.optimizer = keras.optimizers.adam(lr=self.initial_lr)
+        self.compile_options = {"loss": [keras.losses.mean_squared_error], "metrics": ["mse", "mae"]}
+
+
+class FCN(AbstractModelClass):
+    """
+    A customised model 4 Dense layers (64, 32, 16, window_lead_time), where the last layer is the output layer depending
+    on the window_lead_time parameter.
+    """
+
+    _activation = {"relu": keras.layers.ReLU(), "tanh": keras.layers.Activation("tanh"),
+                   "sigmoid": keras.layers.Activation("sigmoid")}
+    _optimizer = {"adam": keras.optimizers.adam, "sgd": keras.optimizers.SGD}
+    _requirements = ["lr", "beta_1", "beta_2", "epsilon", "decay", "amsgrad", "momentum", "nesterov"]
+
+    def __init__(self, input_shape: list, output_shape: list, activation="relu", optimizer="adam",
+                 layers=1, neurons=10, **kwargs):
+        """
+        Sets model and loss depending on the given arguments.
+
+        :param input_shape: list of input shapes (expect len=1 with shape=(window_hist, station, variables))
+        :param output_shape: list of output shapes (expect len=1 with shape=(window_forecast))
+        """
+
+        assert len(input_shape) == 1
+        assert len(output_shape) == 1
+        super().__init__(input_shape[0], output_shape[0])
+
+        # settings
+        self.activation = self._set_activation(activation)
+        self.optimizer = self._set_optimizer(optimizer, **kwargs)
+        self.layer_configuration = (layers, neurons)
+
+        # apply to model
+        self.set_model()
+        self.set_compile_options()
+        # self.set_custom_objects(loss=self.compile_options['loss'])
+
+    def _set_activation(self, activation):
+        try:
+            return self._activation.get(activation.lower())
+        except KeyError:
+            raise AttributeError(f"Given activation {activation} is not supported in this model class.")
+
+    def _set_optimizer(self, optimizer, **kwargs):
+        try:
+            opt_name = optimizer.lower()
+            opt = self._optimizer.get(opt_name)
+            opt_kwargs = {}
+            if opt_name == "adam":
+                opt_kwargs = select_from_dict(kwargs, ["lr", "beta_1", "beta_2", "epsilon", "decay", "amsgrad"])
+            elif opt_name == "sgd":
+                opt_kwargs = select_from_dict(kwargs, ["lr", "momentum", "decay", "nesterov"])
+            return opt(**opt_kwargs)
+        except KeyError:
+            raise AttributeError(f"Given optimizer {optimizer} is not supported in this model class.")
+
+    def set_model(self):
+        """
+        Build the model.
+        """
+        x_input = keras.layers.Input(shape=self._input_shape)
+        x_in = keras.layers.Flatten()(x_input)
+        n_layer, n_hidden = self.layer_configuration
+        for layer in range(n_layer):
+            x_in = keras.layers.Dense(n_hidden)(x_in)
+            x_in = self.activation(x_in)
+        x_in = keras.layers.Dense(self._output_shape)(x_in)
+        out = self.activation(x_in)
+        self.model = keras.Model(inputs=x_input, outputs=[out])
+
+    def set_compile_options(self):
+        self.compile_options = {"loss": [keras.losses.mean_squared_error], "metrics": ["mse", "mae"]}

mlair/run_modules/model_setup.py

@@ -56,7 +56,6 @@ class ModelSetup(RunEnvironment):
         """Initialise and run model setup."""
         super().__init__()
         self.model = None
-        # path = self.data_store.get("experiment_path")
         exp_name = self.data_store.get("experiment_name")
         path = self.data_store.get("model_path")
         self.scope = "model"
@@ -138,9 +137,10 @@ class ModelSetup(RunEnvironment):
 
     def build_model(self):
         """Build model using input and output shapes from data store."""
-        args_list = ["input_shape", "output_shape"]
-        args = self.data_store.create_args_dict(args_list, self.scope)
+        # args_list = ["input_shape", "output_shape"]
         model = self.data_store.get("model_class")
+        args_list = model.requirements()
+        args = self.data_store.create_args_dict(args_list, self.scope)
         self.model = model(**args)
         self.get_model_settings()