first CNN class try

mlair/model_modules/convolutional_networks.py

+__author__ = "Lukas Leufen"
+__date__ = '2021-02-'
+
+from functools import reduce, partial
+
+from mlair.model_modules import AbstractModelClass
+from mlair.helpers import select_from_dict
+from mlair.model_modules.loss import var_loss, custom_loss
+from mlair.model_modules.advanced_paddings import PadUtils, Padding2D, SymmetricPadding2D
+
+import keras
+
+
+class CNN(AbstractModelClass):
+    _activation = {"relu": keras.layers.ReLU, "tanh": partial(keras.layers.Activation, "tanh"),
+                   "sigmoid": partial(keras.layers.Activation, "sigmoid"),
+                   "linear": partial(keras.layers.Activation, "linear"),
+                   "selu": partial(keras.layers.Activation, "selu")}
+    _initializer = {"selu": keras.initializers.lecun_normal()}
+    _optimizer = {"adam": keras.optimizers.adam}
+    _regularizer = {"l1": keras.regularizers.l1, "l2": keras.regularizers.l2, "l1_l2": keras.regularizers.l1_l2}
+    _requirements = ["lr", "beta_1", "beta_2", "epsilon", "decay", "amsgrad"]
+
+    def __init__(self, input_shape: list, output_shape: list, activation="relu", activation_output="linear",
+                 optimizer="adam", regularizer=None, **kwargs):
+
+        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.activation_name = activation
+        self.activation_output = self._set_activation(activation_output)
+        self.activation_output_name = activation_output
+        self.kernel_initializer = self._initializer.get(activation, "glorot_uniform")
+        self.kernel_regularizer = self._set_regularizer(regularizer, **kwargs)
+        self.optimizer = self._set_optimizer(optimizer, **kwargs)
+
+        # apply to model
+        self.set_model()
+        self.set_compile_options()
+        self.set_custom_objects(loss=custom_loss([keras.losses.mean_squared_error, var_loss]), var_loss=var_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"])
+            return opt(**opt_kwargs)
+        except KeyError:
+            raise AttributeError(f"Given optimizer {optimizer} is not supported in this model class.")
+
+    def _set_regularizer(self, regularizer, **kwargs):
+        if regularizer is None or (isinstance(regularizer, str) and regularizer.lower() == "none"):
+            return None
+        try:
+            reg_name = regularizer.lower()
+            reg = self._regularizer.get(reg_name)
+            reg_kwargs = {}
+            if reg_name in ["l1", "l2"]:
+                reg_kwargs = select_from_dict(kwargs, reg_name, remove_none=True)
+                if reg_name in reg_kwargs:
+                    reg_kwargs["l"] = reg_kwargs.pop(reg_name)
+            elif reg_name == "l1_l2":
+                reg_kwargs = select_from_dict(kwargs, ["l1", "l2"], remove_none=True)
+            return reg(**reg_kwargs)
+        except KeyError:
+            raise AttributeError(f"Given regularizer {regularizer} is not supported in this model class.")
+
+    def set_model(self):
+        """
+        Build the model.
+        """
+        x_input = keras.layers.Input(shape=self._input_shape)
+        kernel = (1, 1)
+        pad_size = PadUtils.get_padding_for_same(kernel)
+        x_in = Padding2D("SymPad2D")(padding=pad_size, name="SymPad")(x_input)
+        x_in = keras.layers.Conv2D(filters=16, kernel_size=kernel,
+                                   kernel_initializer=self.kernel_initializer,
+                                   kernel_regularizer=self.kernel_regularizer)(x_in)
+        x_in = self.activation()(x_in)
+        x_in = keras.layers.Conv2D(filters=32, kernel_size=kernel,
+                                   kernel_initializer=self.kernel_initializer,
+                                   kernel_regularizer=self.kernel_regularizer)(x_in)
+        x_in = self.activation()(x_in)
+        x_in = Padding2D("SymPad2D")(padding=pad_size, name="SymPad")(x_in)
+        x_in = keras.layers.Conv2D(filters=64, kernel_size=kernel,
+                                   kernel_initializer=self.kernel_initializer,
+                                   kernel_regularizer=self.kernel_regularizer)(x_in)
+        x_in = self.activation()(x_in)
+        x_in = keras.layers.Flatten()(x_in)
+        x_in = keras.layers.Dense(64, kernel_initializer=self.kernel_initializer,
+                                  kernel_regularizer=self.kernel_regularizer)(x_in)
+        x_in = self.activation()(x_in)
+        x_in = keras.layers.Dense(16, kernel_initializer=self.kernel_initializer,
+                                  kernel_regularizer=self.kernel_regularizer)(x_in)
+        x_in = self.activation()(x_in)
+        x_in = keras.layers.Dense(self._output_shape)(x_in)
+        out = self.activation_output(name=f"{self.activation_output_name}_output")(x_in)
+        self.model = keras.Model(inputs=x_input, outputs=[out])
+
+    def set_compile_options(self):
+        self.compile_options = {"loss": [custom_loss([keras.losses.mean_squared_error, var_loss])],
+                                "metrics": ["mse", "mae", var_loss]}