Allow for TensorFlow 2

.github/workflows/ci.yml

@@ -21,8 +21,8 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: [3.6, 3.7]
-        
+        python-version: [3.6, 3.7, 3.8]
+
     env:
       UCI_DB: "https://archive.ics.uci.edu/ml/machine-learning-databases"
       PROPUBLICA_GH: "https://raw.githubusercontent.com/propublica/compas-analysis/bafff5da3f2e45eca6c2d5055faad269defd135a"
@@ -37,15 +37,15 @@ jobs:
         uses: actions/setup-python@v2
         with:
           python-version: ${{ matrix.python-version }}
-      
+
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip setuptools wheel
           pip install numpy==1.19.5
           pip install -e '.[all]'
           pip install flake8
           pip list
-          
+
       - name: Download datasets
         run: |
           wget ${UCI_DB}/adult/adult.data -P aif360/data/raw/adult/
@@ -54,29 +54,29 @@ jobs:
           wget ${UCI_DB}/statlog/german/german.data -P aif360/data/raw/german/
           wget ${UCI_DB}/statlog/german/german.doc -P aif360/data/raw/german/
           wget ${PROPUBLICA_GH}/compas-scores-two-years.csv -P aif360/data/raw/compas/
-      
+
       - name: Lint with flake8
         run: |
           # stop the build if there are Python syntax errors or undefined names
           flake8 . --count --select=E9,F63,F7,F82 --show-source --statistics
           # exit-zero treats all errors as warnings. The GitHub editor is 127 chars wide
           flake8 . --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics
-      
+
       - name: Test with pytest
         run: pytest tests
-          
+
   build-r:
     runs-on: ubuntu-16.04
-    
+
     strategy:
       fail-fast: false
       matrix:
-        python-version: [3.6, 3.7]
-    
+        python-version: [3.6, 3.7, 3.8]
+
     steps:
       - name: Check out repo
         uses: actions/checkout@v2
-      
+
       - name: Set up R
         uses: r-lib/actions/setup-r@v1
 
@@ -88,11 +88,11 @@ jobs:
       - name: Install R dependencies
         run: install.packages(c("reticulate", "rstudioapi", "testthat"))
         shell: Rscript {0}
-      
+
       - name: Install Python dependencies
         run: |
           python -m pip install --upgrade pip setuptools wheel
           pip install '.[all]'
-          
+
       - name: Install R package
         run: R CMD INSTALL aif360/aif360-r

README.md

@@ -74,9 +74,9 @@ Supported Python Configurations:
 
 | OS      | Python version |
 | ------- | -------------- |
-| macOS   | 3.6, 3.7  |
-| Ubuntu  | 3.6, 3.7  |
-| Windows | 3.6, 3.7  |
+| macOS   | 3.6, 3.7, 3.8  |
+| Ubuntu  | 3.6, 3.7, 3.8  |
+| Windows | 3.6, 3.7, 3.8  |
 
 ### (Optional) Create a virtual environment
 
@@ -176,10 +176,10 @@ issue here and try the solutions.
 
 #### TensorFlow
 
-See the [Install TensorFlow with pip](https://www.tensorflow.org/install/pip?lang=python3#older-versions-of-tensorflow)
+See the [Install TensorFlow with pip](https://www.tensorflow.org/install/pip)
 page for detailed instructions.
 
-Note: we require `'tensorflow >= 1.13.1, < 2'`.
+Note: we require `'tensorflow >= 1.13.1'`.
 
 Once tensorflow is installed, try re-running:
 

aif360/algorithms/inprocessing/adversarial_debiasing.py

@@ -1,7 +1,7 @@
 import numpy as np
 
 try:
-    import tensorflow as tf
+    import tensorflow.compat.v1 as tf
 except ImportError as error:
     from logging import warning
     warning("{}: AdversarialDebiasing will be unavailable. To install, run:\n"
@@ -82,14 +82,14 @@ def _classifier_model(self, features, features_dim, keep_prob):
         """
         with tf.variable_scope("classifier_model"):
             W1 = tf.get_variable('W1', [features_dim, self.classifier_num_hidden_units],
-                                  initializer=tf.contrib.layers.xavier_initializer(seed=self.seed1))
+                                  initializer=tf.initializers.glorot_uniform(seed=self.seed1))
             b1 = tf.Variable(tf.zeros(shape=[self.classifier_num_hidden_units]), name='b1')
 
             h1 = tf.nn.relu(tf.matmul(features, W1) + b1)
             h1 = tf.nn.dropout(h1, keep_prob=keep_prob, seed=self.seed2)
 
             W2 = tf.get_variable('W2', [self.classifier_num_hidden_units, 1],
-                                 initializer=tf.contrib.layers.xavier_initializer(seed=self.seed3))
+                                 initializer=tf.initializers.glorot_uniform(seed=self.seed3))
             b2 = tf.Variable(tf.zeros(shape=[1]), name='b2')
 
             pred_logit = tf.matmul(h1, W2) + b2
@@ -105,7 +105,7 @@ def _adversary_model(self, pred_logits, true_labels):
             s = tf.sigmoid((1 + tf.abs(c)) * pred_logits)
 
             W2 = tf.get_variable('W2', [3, 1],
-                                 initializer=tf.contrib.layers.xavier_initializer(seed=self.seed4))
+                                 initializer=tf.initializers.glorot_uniform(seed=self.seed4))
             b2 = tf.Variable(tf.zeros(shape=[1]), name='b2')
 
             pred_protected_attribute_logit = tf.matmul(tf.concat([s, s * true_labels, s * (1.0 - true_labels)], axis=1), W2) + b2
@@ -123,6 +123,11 @@ def fit(self, dataset):
         Returns:
             AdversarialDebiasing: Returns self.
         """
+        if tf.executing_eagerly():
+            raise RuntimeError("AdversarialDebiasing does not work in eager "
+                    "execution mode. To fix, add `tf.disable_eager_execution()`"
+                    " to the top of the calling script.")
+
         if self.seed is not None:
             np.random.seed(self.seed)
         ii32 = np.iinfo(np.int32)

aif360/sklearn/inprocessing/adversarial_debiasing.py

@@ -4,7 +4,7 @@
 from sklearn.preprocessing import LabelEncoder
 from sklearn.utils import check_random_state
 from sklearn.utils.validation import check_is_fitted
-import tensorflow as tf
+import tensorflow.compat.v1 as tf
 
 from aif360.sklearn.utils import check_inputs, check_groups
 
@@ -91,6 +91,11 @@ def fit(self, X, y):
         Returns:
             self
         """
+        if tf.executing_eagerly():
+            raise RuntimeError("AdversarialDebiasing does not work in eager "
+                    "execution mode. To fix, add `tf.disable_eager_execution()`"
+                    " to the top of the calling script.")
+
         X, y, _ = check_inputs(X, y)
         rng = check_random_state(self.random_state)
         ii32 = np.iinfo(np.int32)

examples/demo_adversarial_debiasing.ipynb

@@ -44,7 +44,8 @@
     "from IPython.display import Markdown, display\n",
     "import matplotlib.pyplot as plt\n",
     "\n",
-    "import tensorflow as tf"
+    "import tensorflow.compat.v1 as tf\n",
+    "tf.disable_eager_execution()"
    ]
   },
   {

examples/sklearn/demo_new_features.ipynb

@@ -17,7 +17,8 @@
     "import matplotlib.pyplot as plt\n",
     "import numpy as np\n",
     "import pandas as pd\n",
-    "import tensorflow as tf\n",
+    "import tensorflow.compat.v1 as tf\n",
+    "tf.disable_eager_execution()\n",
     "tf.logging.set_verbosity(tf.logging.ERROR)\n",
     "\n",
     "from sklearn.compose import make_column_transformer\n",

setup.py

@@ -15,7 +15,7 @@
 
 extras = {
     'OptimPreproc': ['cvxpy>=1.0'],
-    'AdversarialDebiasing': ['tensorflow>=1.13.1,<2'],
+    'AdversarialDebiasing': ['tensorflow>=1.13.1'],
     'DisparateImpactRemover': ['BlackBoxAuditing'],
     'LIME': ['lime'],
     'ART': ['adversarial-robustness-toolbox>=1.0.0'],

tests/sklearn/test_adversarial_debiasing.py

@@ -1,7 +1,8 @@
 import numpy as np
 from sklearn.model_selection import GridSearchCV
 from sklearn.metrics import accuracy_score
-import tensorflow as tf
+import tensorflow.compat.v1 as tf
+tf.disable_eager_execution()
 
 from aif360.datasets import AdultDataset
 from aif360.sklearn.datasets import fetch_adult