pg-sui 0.2.3__py3-none-any.whl → 1.6.14.dev9__py3-none-any.whl

pgsui/impute/unsupervised/models/in_development/cnn_model.py

@@ -1,486 +0,0 @@
-import logging
-import os
-import sys
-import warnings
-import math
-
-# Import tensorflow with reduced warnings.
-os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"
-logging.getLogger("tensorflow").disabled = True
-warnings.filterwarnings("ignore", category=UserWarning)
-
-import numpy as np
-import pandas as pd
-import tensorflow as tf
-
-# Disable can't find cuda .dll errors. Also turns of GPU support.
-tf.config.set_visible_devices([], "GPU")
-
-from tensorflow.python.util import deprecation
-
-# Disable warnings and info logs.
-tf.compat.v1.logging.set_verbosity(tf.compat.v1.logging.ERROR)
-tf.get_logger().setLevel(logging.ERROR)
-
-
-# Monkey patching deprecation utils to supress warnings.
-# noinspection PyUnusedLocal
-def deprecated(
-    date, instructions, warn_once=True
-):  # pylint: disable=unused-argument
-    def deprecated_wrapper(func):
-        return func
-
-    return deprecated_wrapper
-
-
-deprecation.deprecated = deprecated
-
-from tensorflow.keras.layers import (
-    Dropout,
-    Dense,
-    Reshape,
-    Activation,
-    Flatten,
-    BatchNormalization,
-    LeakyReLU,
-    PReLU,
-)
-
-from tensorflow.keras.regularizers import l1_l2
-
-# Custom Modules
-try:
-    from ...neural_network_methods import NeuralNetworkMethods
-except (ModuleNotFoundError, ValueError, ImportError):
-    from impute.unsupervised.neural_network_methods import NeuralNetworkMethods
-
-
-class SoftOrdering1DCNN(tf.keras.Model):
-    def __init__(
-        self,
-        y=None,
-        output_shape=None,
-        weights_initializer="glorot_normal",
-        hidden_layer_sizes="midpoint",
-        num_hidden_layers=1,
-        hidden_activation="elu",
-        l1_penalty=1e-6,
-        l2_penalty=1e-6,
-        dropout_rate=0.2,
-        num_classes=4,
-        sample_weight=None,
-        batch_size=32,
-        missing_mask=None,
-        activation=None,
-        channel_increase_rate=2,
-        initial_hidden_size=2048,
-        num_groups=256,
-    ):
-        super(SoftOrdering1DCNN, self).__init__()
-
-        self._y = y
-        self._missing_mask = missing_mask
-        self._sample_weight = sample_weight
-        self._batch_idx = 0
-        self._batch_size = batch_size
-        self.output_activation = activation
-        self.sample_weight = sample_weight
-
-        self.nn_ = NeuralNetworkMethods()
-        self.binary_accuracy = self.nn_.make_masked_binary_accuracy(
-            is_vae=True
-        )
-
-        self.total_loss_tracker = tf.keras.metrics.Mean(name="loss")
-        self.reconstruction_loss_tracker = tf.keras.metrics.Mean(
-            name="reconstruction_loss"
-        )
-        self.kl_loss_tracker = tf.keras.metrics.Mean(name="kl_loss")
-        self.accuracy_tracker = tf.keras.metrics.Mean(name="accuracy")
-
-        # y_train[1] dimension.
-        self.n_features = output_shape * num_classes
-
-        self.weights_initializer = weights_initializer
-        self.hidden_layer_sizes = hidden_layer_sizes
-        self.num_hidden_layers = num_hidden_layers
-        self.hidden_activation = hidden_activation
-        self.l1_penalty = l1_penalty
-        self.l2_penalty = l2_penalty
-        self.dropout_rate = dropout_rate
-        self.num_classes = num_classes
-        self.channel_increase_rate = channel_increase_rate
-        self.initial_hidden_size = initial_hidden_size
-        self.channel_size1 = num_groups
-        self.channel_size2 = num_groups * 2
-        self.channel_size3 = num_groups * 2
-
-        nn = NeuralNetworkMethods()
-
-        # hidden_layer_sizes = nn.validate_hidden_layers(
-        #     self.hidden_layer_sizes, self.num_hidden_layers
-        # )
-
-        # hidden_layer_sizes = nn.get_hidden_layer_sizes(
-        #     self.n_features, self.n_components, hidden_layer_sizes, vae=True
-        # )
-
-        # hidden_layer_sizes = [h * self.num_classes for h in hidden_layer_sizes]
-
-        if self.l1_penalty == 0.0 and self.l2_penalty == 0.0:
-            kernel_regularizer = None
-        else:
-            kernel_regularizer = l1_l2(self.l1_penalty, self.l2_penalty)
-
-        kernel_initializer = self.weights_initializer
-
-        if self.hidden_activation.lower() == "leaky_relu":
-            activation = LeakyReLU(alpha=0.01)
-
-        elif self.hidden_activation.lower() == "prelu":
-            activation = PReLU()
-
-        elif self.hidden_activation.lower() == "selu":
-            activation = "selu"
-            kernel_initializer = "lecun_normal"
-
-        else:
-            activation = self.hidden_activation
-
-        if num_hidden_layers > 5:
-            raise ValueError(
-                f"The maximum number of hidden layers is 5, but got "
-                f"{num_hidden_layers}"
-            )
-
-        hidden_size = initial_hidden_size
-        if self.n_features >= hidden_size:
-            scaling_factor = int(math.ceil(self.n_features / hidden_size)) * 2
-            hidden_size *= num_groups * int(
-                math.ceil((scaling_factor / num_groups))
-            )
-        else:
-            # If hidden_size is close in number to n_features
-            if abs(hidden_size - self.n_features) <= (hidden_size // 2):
-                hidden_size *= 2
-
-        # Model adapted from: https://medium.com/spikelab/convolutional-neural-networks-on-tabular-datasets-part-1-4abdd67795b6
-
-        signal_size1 = hidden_size // num_groups
-        signal_size2 = signal_size1 // 2
-        signal_size3 = signal_size1 // 4 * self.channel_size3
-
-        self.signal_size1 = signal_size1
-        self.signal_size2 = signal_size2
-        self.signal_size3 = signal_size3
-
-        self.batch_norm1 = BatchNormalization()
-        self.dropout1 = Dropout(self.dropout_rate)
-        self.dense1 = Dense(
-            hidden_size,
-            input_shape=(self.n_features,),
-            activation=hidden_activation,
-            kernel_initializer=kernel_initializer,
-        )
-
-        self.rshp = Reshape((num_groups, signal_size1))
-
-        self.batch_norm_c1 = BatchNormalization()
-        self.conv1 = tf.keras.layers.Conv1D(
-            self.channel_size1 * self.channel_increase_rate,
-            kernel_size=5,
-            stride=1,
-            padding=2,
-            groups=signal_size1,
-            kernel_initializer=kernel_initializer,
-            activation=hidden_activation,
-        )
-
-        self.avg_po_c1 = tf.keras.layers.AveragePooling1D(
-            pool_size=4, padding="valid"
-        )
-
-        self.batch_norm_c2 = BatchNormalization()
-        self.dropout_c2 = Dropout(self.dropout_rate)
-        self.conv2 = tf.keras.layers.Conv1D(
-            self.channel_size2,
-            kernel_size=3,
-            stride=1,
-            padding=1,
-            kernel_initializer=kernel_initializer,
-            activation=hidden_activation,
-        )
-
-        self.batch_norm_c3 = BatchNormalization()
-        self.dropout_c3 = Dropout(self.dropout_rate)
-        self.conv3 = tf.keras.layers.Conv1D(
-            self.channel_size2,
-            kernel_size=3,
-            stride=1,
-            padding=1,
-            kernel_initializer=kernel_initializer,
-            activation=hidden_activation,
-        )
-
-        self.batch_norm_c4 = BatchNormalization()
-        self.dropout_c4 = Dropout(self.dropout_rate)
-        self.conv4 = tf.keras.layers.Conv1D(
-            self.channel_size2,
-            kernel_size=5,
-            stride=1,
-            padding=2,
-            groups=signal_size1,
-            kernel_initializer=kernel_initializer,
-            activation=None,
-        )
-
-        self.act_c4 = Activation(hidden_activation)
-
-        self.max_po_c4 = tf.keras.layers.MaxPooling1D(
-            pool_size=4, stride=2, padding=1
-        )
-
-        self.flatten = Flatten()
-
-        self.batch_norm2 = BatchNormalization()
-        self.dropout2 = Dropout(self.dropout_rate)
-        self.dense2 = Dense(
-            self.n_features, kernel_initializer=kernel_initializer
-        )
-        self.rshp2 = Reshape((output_shape, num_classes))
-        self.act2 = Activation(activation)
-
-    def call(self, inputs, training=None):
-        """Call the model on a particular input.
-
-        Args:
-            input (tf.Tensor): Input tensor. Must be one-hot encoded.
-
-        Returns:
-            tf.Tensor: Output predictions. Will be one-hot encoded.
-        """
-        x = self.dense1(inputs)
-        x = self.batch_norm1(x, training=training)
-        x = self.dropout1(x, training=training)
-        x = self.rshp(x)
-        x = self.conv1(x)
-        x = self.batch_norm_c1(x, training=training)
-        x = self.avg_po_c1(x)
-        x = self.conv2(x)
-        x = self.batch_norm(x, training=training)
-        x = self.dropout_c2(x, training=training)
-        x_s = x
-        x = self.conv3(x)
-        x = self.batch_norm_c3(x, training=training)
-        x = self.dropout(x, training=training)
-        x = self.conv4(x)
-        x = self.batch_norm_c4(x, training=training)
-        x += x_s
-        x = self.act_c4(x)
-        x = self.max_po_c4(x)
-        x = self.dropout1(x)
-        x = self.rshp(x)
-        x = self.batch_norm_c1(x)
-        x = self.conv1(x)
-        x = self.avg_po_c1(x)
-        x = self.flatten(x)
-        x = self.dense2(x)
-        x = self.batch_norm2(x, training=training)
-        x = self.dropout2(x, training=training)
-        x = self.rshp2(x)
-        return self.act2(x)
-
-    def model(self):
-        """Here so that mymodel.model().summary() can be called for debugging."""
-        x = tf.keras.Input(shape=(self.n_features * self.num_classes,))
-        return tf.keras.Model(inputs=[x], outputs=self.call(x))
-
-    def set_model_outputs(self):
-        x = tf.keras.Input(shape=(self.n_features * self.num_classes,))
-        model = tf.keras.Model(inputs=[x], outputs=self.call(x))
-        self.outputs = model.outputs
-
-    @property
-    def metrics(self):
-        return [
-            self.total_loss_tracker,
-            self.reconstruction_loss_tracker,
-            self.kl_loss_tracker,
-            self.accuracy_tracker,
-        ]
-
-    @tf.function
-    def train_step(self, data):
-        # if isinstance(data, tuple):
-        #     if len(data) == 2:
-        #         x, y = data
-        #         sample_weight = None
-        #     else:
-        #         x, y, sample_weight = data
-        # else:
-        #     raise TypeError("Target y must be supplied to fit for this model.")
-
-        # Set in the UBPCallbacks() callback.
-        y = self._y
-
-        (
-            y,
-            y_true,
-            sample_weight,
-            missing_mask,
-            batch_start,
-            batch_end,
-        ) = self.nn_.prepare_training_batches(
-            y,
-            y,
-            self._batch_size,
-            self._batch_idx,
-            True,
-            self.n_components,
-            self._sample_weight,
-            self._missing_mask,
-            ubp=False,
-        )
-
-        if sample_weight is not None:
-            sample_weight_masked = tf.convert_to_tensor(
-                sample_weight[~missing_mask], dtype=tf.float32
-            )
-        else:
-            sample_weight_masked = None
-
-        y_true_masked = tf.boolean_mask(
-            tf.convert_to_tensor(y_true, dtype=tf.float32),
-            tf.reduce_any(tf.not_equal(y_true, -1), axis=2),
-        )
-
-        with tf.GradientTape() as tape:
-            reconstruction = self(tf.convert_to_tensor(y), training=True)
-
-            y_pred_masked = tf.boolean_mask(
-                reconstruction, tf.reduce_any(tf.not_equal(y_true, -1), axis=2)
-            )
-
-            # Returns binary crossentropy loss.
-            loss = self.compiled_loss(
-                y_true_masked,
-                y_pred_masked,
-                sample_weight=sample_weight_masked,
-            )
-
-        grads = tape.gradient(loss, self.trainable_weights)
-        self.optimizer.apply_gradients(zip(grads, self.trainable_weights))
-        self.total_loss_tracker.update_state(loss)
-
-        ### NOTE: If you get the error, "'tuple' object has no attribute
-        ### 'rank', then convert y_true to a tensor object."
-        # self.compiled_metrics.update_state(
-        self.accuracy_tracker.update_state(
-            self.binary_accuracy(
-                y_true_masked,
-                y_pred_masked,
-                sample_weight=sample_weight_masked,
-            )
-        )
-
-        return {
-            "loss": self.total_loss_tracker.result(),
-            "accuracy": self.accuracy_tracker.result(),
-        }
-
-    @tf.function
-    def test_step(self, data):
-        if isinstance(data, tuple):
-            if len(data) == 2:
-                x, y = data
-                sample_weight = None
-            else:
-                x, y, sample_weight = data
-        else:
-            raise TypeError("Target y must be supplied to fit in this model.")
-
-        if sample_weight is not None:
-            sample_weight_masked = tf.boolean_mask(
-                tf.convert_to_tensor(sample_weight),
-                tf.reduce_any(tf.not_equal(y, -1), axis=2),
-            )
-        else:
-            sample_weight_masked = None
-
-        reconstruction, z_mean, z_log_var, z = self(x, training=False)
-        reconstruction_loss = self.compiled_loss(
-            y,
-            reconstruction,
-            sample_weight=sample_weight_masked,
-        )
-
-        # Includes KL Divergence Loss.
-        regularization_loss = sum(self.losses)
-
-        total_loss = reconstruction_loss + regularization_loss
-
-        self.accuracy_tracker.update_state(
-            self.cateogrical_accuracy(
-                y,
-                reconstruction,
-                sample_weight=sample_weight_masked,
-            )
-        )
-
-        self.total_loss_tracker.update_state(total_loss)
-        self.reconstruction_loss_tracker.update_state(reconstruction_loss)
-        self.kl_loss_tracker.update_state(regularization_loss)
-
-        return {
-            "loss": self.total_loss_tracker.result(),
-            "reconstruction_loss": self.reconstruction_loss_tracker.result(),
-            "kl_loss": self.kl_loss_tracker.result(),
-            "accuracy": self.accuracy_tracker.result(),
-        }
-
-    @property
-    def batch_size(self):
-        """Batch (=step) size per epoch."""
-        return self._batch_size
-
-    @property
-    def batch_idx(self):
-        """Current batch (=step) index."""
-        return self._batch_idx
-
-    @property
-    def y(self):
-        return self._y
-
-    @property
-    def missing_mask(self):
-        return self._missing_mask
-
-    @property
-    def sample_weight(self):
-        return self._sample_weight
-
-    @batch_size.setter
-    def batch_size(self, value):
-        """Set batch_size parameter."""
-        self._batch_size = int(value)
-
-    @batch_idx.setter
-    def batch_idx(self, value):
-        """Set current batch (=step) index."""
-        self._batch_idx = int(value)
-
-    @y.setter
-    def y(self, value):
-        """Set y after each epoch."""
-        self._y = value
-
-    @missing_mask.setter
-    def missing_mask(self, value):
-        """Set y after each epoch."""
-        self._missing_mask = value
-
-    @sample_weight.setter
-    def sample_weight(self, value):
-        self._sample_weight = value