Models: Update models to keras 3

ceruleo/models/keras/catalog/MSWRLRCN.py

@@ -1,9 +1,8 @@
 from typing import Tuple
 import tensorflow as tf
-from tensorflow.python.keras.layers.pooling import GlobalAveragePooling1D, MaxPool1D
 from ceruleo.models.keras.layers import ExpandDimension, ResidualShrinkageBlock
-from tensorflow.keras import Input, Model, Sequential
-from tensorflow.keras.layers import (
+from keras import Input, Model, Sequential
+from keras.layers import (
 
     BatchNormalization,
 
@@ -14,6 +13,8 @@
     LSTM,
     Bidirectional,
     Concatenate,
+    GlobalAveragePooling1D, 
+    MaxPooling1D
 
 )
 from typing import Tuple
@@ -32,7 +33,7 @@ def ConvBlock(n_filters: int, kernel_size: int):
                 Conv1D(n_filters, kernel_size, padding='same'),
                 BatchNormalization(),
                 ReLU(),
-                MaxPool1D(2),
+                MaxPooling1D(2),
             ]
         )
 

ceruleo/models/keras/catalog/XCM.py

@@ -3,8 +3,8 @@
 import tensorflow as tf
 
 from ceruleo.models.keras.layers import ExpandDimension, RemoveDimension
-from tensorflow.keras import Input, Model
-from tensorflow.keras.layers import (
+from keras import Input, Model
+from keras.layers import (
     Activation,
     BatchNormalization,
     Concatenate,
@@ -96,7 +96,7 @@ def XCM(input_shape: Tuple[int, int], *, n_filters: int = 128, filter_window: in
 
     model = Model(
         inputs=[model_input],
-        outputs=[output],
+        outputs=output,
     )
     return model,  (model_fisrt_conv1d, model_fisrt_conv2d, model_regression)
 

ceruleo/models/keras/layers.py

@@ -1,28 +1,36 @@
-import typing
-import warnings
-from typing import Tuple
+
 
 import numpy as np
 import tensorflow as tf
-from tensorflow.keras import Sequential
-from tensorflow.keras import backend as K
-from tensorflow.keras import regularizers
-from tensorflow.keras.layers import (Activation, BatchNormalization, Conv2D,
-                                     Dense, Flatten, Lambda, Layer, Permute,
-                                     Reshape)
+import keras
+from keras import Sequential
+from keras import backend as K
+from keras import regularizers
+from keras.layers import (
+    Activation,
+    BatchNormalization,
+    Conv2D,
+    GlobalAveragePooling1D,
+    Dense,
+    Flatten,
+    Lambda,
+    Layer,
+    Permute,
+    maximum,
+    Reshape,
+)
 from tensorflow.python.framework import tensor_shape
-from tensorflow.python.keras.layers.pooling import GlobalAveragePooling1D
 
 
 def ExpandDimension(dim: int = -1):
-    return Lambda(lambda x: K.expand_dims(x, dim))
+    return Lambda(lambda x: keras.ops.expand_dims(x, dim))
 
 
 def RemoveDimension(axis=0):
-    return Lambda(lambda x: K.squeeze(x, axis=axis))
+    return Lambda(lambda x: keras.ops.squeeze(x, axis=axis))
 
 
-class ConcreteDropout(tf.keras.layers.Layer):
+class ConcreteDropout(Layer):
     """
     Concrete Dropout layer class from https://arxiv.org/abs/1705.07832.
     Dropout Feature Ranking for Deep Learning Models
@@ -50,9 +58,8 @@ def __init__(
         init_max=0.9,
         name=None,
         training=True,
-        **kwargs
+        **kwargs,
     ):
-
         super(ConcreteDropout, self).__init__(name=name, **kwargs)
         assert init_min <= init_max, "init_min must be lower or equal to init_max."
 
@@ -96,7 +103,6 @@ def concrete_dropout(self, p, x):
         return x
 
     def call(self, inputs, training=True):
-
         p = K.sigmoid(self.p_logit)
 
         dropout_regularizer = p * K.log(p)
@@ -110,11 +116,12 @@ def call(self, inputs, training=True):
         return x
 
 
-class ResidualShrinkageBlock(tf.keras.layers.Layer):
+class ResidualShrinkageBlock(Layer):
     """
     ResidualShrinkageBlock
 
     """
+
     def build(self, input_shape):
         self.blocks = []
         for i in range(2):
@@ -171,63 +178,57 @@ def call(self, inputs):
 
         zeros = sub - sub
 
-        n_sub = tf.keras.layers.maximum([sub, zeros])
+        n_sub = maximum([sub, zeros])
 
-        residual = tf.keras.backend.sign(residual) * n_sub
+        residual = keras.ops.sign(residual) * n_sub
         residual = RemoveDimension(3)(residual)
         return residual + inputs
 
 
-
-
 class ZeroWeights(tf.keras.constraints.Constraint):
-
-
-  def __init__(self, l1:float):
-    self.l1 = l1
+    def __init__(self, l1: float):
+        self.l1 = l1
 
-  def __call__(self, w):
-
-    return (tf.math.multiply(w, tf.cast(tf.abs(w) > self.l1, tf.float32)) )
+    def __call__(self, w):
+        return tf.math.multiply(w, tf.cast(tf.abs(w) > self.l1, tf.float32))
 
-  def get_config(self):
-    return {'l1': self.l1}
+    def get_config(self):
+        return {"l1": self.l1}
 
 
 class LASSOLayer(Layer):
-    """ 
+    """
     LASSO Layer
 
     Parameters:
         l1: L1 regularization parameter
     """
-    def __init__(self, l1:float):
+
+    def __init__(self, l1: float):
         super(LASSOLayer, self).__init__()
         self.l1 = l1
         self.kernel_regularizer = regularizers.L1(l1)
 
-
     def build(self, input_shape):
         W_size = np.prod(input_shape[1:])
         self.w = self.add_weight(
-            shape=(W_size, ),
+            shape=(W_size,),
             initializer="random_normal",
             trainable=True,
             regularizer=self.kernel_regularizer,
-            constraint=ZeroWeights(self.l1)
+            constraint=ZeroWeights(self.l1),
         )
-
-
+
         self.input_reshape = Reshape((W_size,))
         self.output_reshape = Reshape(input_shape[1:])
-
-
 
     def call(self, inputs):
         x = self.input_reshape(inputs)
-
-
-        x = tf.math.multiply(self.w, x) 
-
-        self.add_metric(tf.math.reduce_sum(tf.cast(tf.abs(self.w) > 0, tf.float32)), name="Number of features")
+
+        x = tf.math.multiply(self.w, x)
+
+        self.add_metric(
+            tf.math.reduce_sum(tf.cast(tf.abs(self.w) > 0, tf.float32)),
+            name="Number of features",
+        )
         return self.output_reshape(x)

docs/dataset/analysis/Sensor Validation.ipynb

@@ -1131,7 +1131,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.11.7"
+   "version": "3.11.9"
   }
  },
  "nbformat": 4,

pyproject.toml

@@ -42,7 +42,7 @@ Homepage = "https://github.com/lucianolorenti/ceruleo"
 Documentation = "https://lucianolorenti.github.io/ceruleo/"
 
 [project.optional-dependencies]
-tensorflow = ["tensorflow >= 2.5", "keras < 3"]
+tensorflow = ["tensorflow >= 2.5"]
 
 test = [
     "pytest",
@@ -62,4 +62,4 @@ doc = [
     "sphinxcontrib.bibtex",
     "mkdocs-matplotlib",
 
-]
+]

tests/test_models.py

@@ -3,6 +3,13 @@
 import numpy as np
 import pandas as pd
 import tensorflow as tf
+from numpy.random import seed
+from sklearn.linear_model import LinearRegression
+from sklearn.pipeline import make_pipeline
+from tensorflow.keras import Input, Model
+from tensorflow.keras.layers import Dense, Flatten
+from xgboost import XGBRegressor
+
 from ceruleo.dataset.ts_dataset import AbstractPDMDataset
 from ceruleo.iterators.iterators import WindowedDatasetIterator
 from ceruleo.iterators.shufflers import AllShuffled
@@ -12,7 +19,6 @@
 from ceruleo.models.keras.catalog.CNLSTM import CNLSTM
 from ceruleo.models.keras.catalog.InceptionTime import InceptionTime
 from ceruleo.models.keras.catalog.MSWRLRCN import MSWRLRCN
-from ceruleo.models.keras.catalog.MVCNN import MVCNN
 from ceruleo.models.keras.catalog.MultiScaleConvolutional import (
     MultiScaleConvolutionalModel,
 )
@@ -36,12 +42,6 @@
 from ceruleo.transformation import Transformer
 from ceruleo.transformation.features.scalers import MinMaxScaler
 from ceruleo.transformation.features.selection import ByNameFeatureSelector
-from numpy.random import seed
-from sklearn.linear_model import LinearRegression
-from sklearn.pipeline import make_pipeline
-from tensorflow.keras import Input, Model
-from tensorflow.keras.layers import Dense, Flatten
-from xgboost import XGBRegressor
 
 seed(1)