Repository refactoring and ghub workflow (#3)

* Repo refactor, update of readme and tuto - source code in airobas/ folder - update of dependencies in pyproject.toml - basic github workflow to install lib and dependencies and run pipeline tests - update of read me content
airbus · Aug 1, 2024 · 420d573 · 420d573
1 parent 8da8625
commit 420d573
Show file tree

Hide file tree

Showing 42 changed files with 371 additions and 228 deletions.
diff --git a/.github/workflows/python-package.yml b/.github/workflows/python-package.yml
@@ -0,0 +1,29 @@
+name: Python package
+
+on: [push, pull_request]
+
+jobs:
+  build:
+
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.9", "3.10"]
+
+    steps:
+    - uses: actions/checkout@v3
+    - name: Set up Python ${{ matrix.python-version }}
+      uses: actions/setup-python@v5
+      with:
+        python-version: ${{ matrix.python-version }}
+        cache: "pip"
+        cache-dependency-path: pyproject.toml
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install .
+        pip install onnxruntime
+        pip install pytest
+    - name: Test with pytest
+      run: |
+        pytest tests/
diff --git a/.gitignore b/.gitignore
@@ -0,0 +1,19 @@
+# Mac
+.DS_Store
+
+# IDE
+.*/
+
+# Python
+__pycache__/
+.python-version
+
+# Temp files
+*.tmp
+
+# Build files
+build/
+dist/
+# don't update locally created images and dump file
+tutorials/dump.txt
+tutorials/rosenbrock_images/*.png
diff --git a/README.md b/README.md
@@ -2,7 +2,8 @@
 
 ## Introduction
 
-**What is Airobas?** `A(i)robas` is a library that combines verification tool for neural network into a pipeline.
+**What is Airobas?** `A(i)robas` is a library that allows the combinaison of verification tools for neural network into a pipeline.
+
 It provides functionalities to:
 
 – load input data and models,
@@ -11,39 +12,33 @@ It provides functionalities to:
 to be as generic as possible and highly parametrizable regarding property
 definition,
 
-– design customized verification pipeline,
+– design a customized verification pipeline,
 
-– run verification with a default set of empirical and formal methods e.g., at-
-tacks using the [cleverhans](https://github.com/cleverhans-lab/cleverhans) library, incomplete verification using LiRPA-based
-functionalities implemented in [decomon](https://github.com/airbus/decomon) etc.,
+– run verification with a default set of empirical and formal methods e.g., adversarial attacks using the [cleverhans](https://github.com/cleverhans-lab/cleverhans) library, incomplete verification using LiRPA-based
+functionalities implemented in the Airbus open-source library [decomon](https://github.com/airbus/decomon) etc.,
 
-– append new verification functions or link the code to open-source libraries
-of one’s choice
+– append new verification functions or link the code to the open-source libraries of one’s choice
 
-We believe that Airobas is a complementary tool to existing libraries for the certification of neural networks.
-This open source code is a support to our publication ["Surrogate Neural Networks Local Stability for
-Aircraft Predictive Maintenance"](https://arxiv.org/abs/2401.06821)
+We believe that Airobas is a complementary tool to existing libraries for the robustness verification of neural networks.
+This open source code is a complement to the publication ["Surrogate Neural Networks Local Stability for
+Aircraft Predictive Maintenance"](https://arxiv.org/abs/2401.06821) which was peer-reviewed and accepted at the 29th International Conference on Formal Methods for Industrial Critical Systems ([FMICS2024](https://fmics.inria.fr/2024/)).
 
 **Complete Verification Pipeline for Stability**
 
-The state-of-the-art property verification of neural networks is currently divided into three
-families of methods. They all take as input test points and a trained model and return
-for every test point if the property is violated or verified.
-- Firstly, there are the ’no/maybe’ methods (block A), such as [adversarial attacks](https://github.com/cleverhans-lab/cleverhans). Essentially, these methods rely on the search for counterexamples that would
-contradict the stability property. If no counterexample is found, no conclusion can
-be drawn.
-- Secondly, there are the ’yes/maybe’ (block B) methods, such as the affine bounds
-implemented in the [Decomon library](https://github.com/airbus/decomon). Here, the outputs of a network are
-bounded. If the bounds found are within the stability bounds, then the property
-is verified. If the bounds are too loose and exceed the required stability interval,
-no conclusion can be drawn. So far, these methods called incomplete methods are
-partial in the sense that they do not provide an absolute solution to the question of
-a network’s stability but are generally fast.
-- The last family of methods corresponds to the ’yes/no’ complete methods (block C). Examples include [SMT](https://github.com/NeuralNetworkVerification/Marabou) or [MILP](https://gurobi-machinelearning.readthedocs.io/en/stable/index.html) solvers, which provide an exact answer
-at the cost of significant computation time.
+Verification techniques for neural networks take test points, a trained model and a given property (robustness, stability, monotonicity etc.) as input and return for every test point an assessment of whether or not the property is violated or verified.
+
+The current state-of-the-art mainly encaompasses three families of methods:
+
+- The ’no/maybe’ methods (Family of techniques 'A' in paper): e.g. [adversarial attacks](https://github.com/cleverhans-lab/cleverhans). These methods essentially rely on the search for counterexamples that contradict the targeted property. If no counterexample is found, no conclusion can be drawn on the model output w.r.t to the targeted property.
+
+- The ’yes/maybe’ methods (Family of techniques 'B' in paper): e.g. the affine bounds generation methods implemented in the [decomon library](https://github.com/airbus/decomon). These techniques intend to bound the output values of a network. If the derived bounds respect the property, it is verified. If the bounds exceed it, no conclusion can be drawn. It can then either mean that the propety is violated or that the derived bounds are too loose. These methods are commonly refered to as 'incomplete' (since they can guarantee the property is respected but not that it is violated). They are however generally faster than 'complete' methods (see next).
+
+- The ’yes/no’ methods (Family of techniques 'C' in paper): e.g., [SMT](https://github.com/NeuralNetworkVerification/Marabou), [MILP](https://gurobi-machinelearning.readthedocs.io/en/stable/index.html) solvers etc. They can guarantee that the neural network output resepct or not the targeted property, at the cost of significant computation time. They are commonly refered to as 'complete' techniques.
+
+Here is an example of NN verification pipeline:
 
 <div align="center">
-    <img src="https://github.com/airbus/Airobas/blob/main/docs/pipeline_plusnumbers_plusletters.png" width="75%" alt="pipeline" align="center" />
+    <img src="https://github.com/airbus/Airobas/blob/main/docs/pipeline.png" width="75%" alt="pipeline" align="center" />
 </div>
 
 
@@ -53,15 +48,11 @@ Quick version:
 ```shell
 pip install .
 ```
-For more details, see the [online documentation](https://airbus.github.io/airobas/main/install).
 
+## Examples
 
+Notebooks providing examples of the use of the library are available in the "tutorials" folder:
 
+- Use case "Aircraft braking distance estimation" [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/airbus/Airobas/blob/main/tutorials/braking_distance_estimation_verification.ipynb)
 
-## Documentation
-
-The latest documentation is available [online](https://airbus.github.io/airobas).
-
-## Examples
-
-Some educational notebooks will be available *soon*.
+- Use case "NN surrogate for Rosenbrock function" [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/airbus/Airobas/blob/main/tutorials/rosenbrock_verification.ipynb)
diff --git a/blocks_hub/__init__.py → airobas/__init__.py b/blocks_hub/__init__.py → airobas/__init__.py
diff --git a/...ib/commons/bounds_propagation/__init__.py → airobas/blocks_hub/__init__.py b/...ib/commons/bounds_propagation/__init__.py → airobas/blocks_hub/__init__.py
diff --git a/blocks_hub/adv_block.py → airobas/blocks_hub/adv_block.py b/blocks_hub/adv_block.py → airobas/blocks_hub/adv_block.py
@@ -7,8 +7,8 @@
 from cleverhans.tf2.attacks.fast_gradient_method import fast_gradient_method
 from cleverhans.tf2.attacks.projected_gradient_descent import projected_gradient_descent
 
-from blocks_hub.meta_block import MetaBlock
-from verif_pipeline import (
+from airobas.blocks_hub.meta_block import MetaBlock
+from airobas.verif_pipeline import (
     BlockVerif,
     BlockVerifOutput,
     DataContainer,

diff --git a/blocks_hub/decomon_block.py → airobas/blocks_hub/decomon_block.py b/blocks_hub/decomon_block.py → airobas/blocks_hub/decomon_block.py
@@ -3,7 +3,7 @@
 import numpy as np
 from decomon.models import clone
 
-from verif_pipeline import BlockVerif, BlockVerifOutput, StatusVerif
+from airobas.verif_pipeline import BlockVerif, BlockVerifOutput, StatusVerif
 
 
 def check_SB_unsat(y_pred_min, y_pred_max, y_min, y_max):

diff --git a/blocks_hub/gml_mip_block.py → airobas/blocks_hub/gml_mip_block.py b/blocks_hub/gml_mip_block.py → airobas/blocks_hub/gml_mip_block.py
@@ -6,31 +6,27 @@
 import numpy as np
 from gurobi_ml import add_predictor_constr
 
-from blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_utils import (
-    create_object_bounds,
-)
-from blocks_hub.mip_blocks_lib.commons.formula import (
+from airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_utils import create_object_bounds
+from airobas.blocks_hub.mip_blocks_lib.commons.formula import (
     GT,
     LT,
     NAryConjFormula,
     StateCoordinate,
     VarConstConstraint,
 )
-from blocks_hub.mip_blocks_lib.commons.neural_network import (
+from airobas.blocks_hub.mip_blocks_lib.commons.neural_network import (
     InputBoundsNeuralNetwork,
     NeuralNetwork,
 )
-from blocks_hub.mip_blocks_lib.commons.parameters import (
+from airobas.blocks_hub.mip_blocks_lib.commons.parameters import (
     ParamsBoundComputation,
     ParamsBoundComputationEnum,
 )
-from blocks_hub.mip_blocks_lib.commons.utilities.evaluate import evaluate, random_input
-from blocks_hub.mip_blocks_lib.linear.verification_formula.verification_constraints import (
-    get_callback_specific,
+from airobas.blocks_hub.mip_blocks_lib.commons.utilities.evaluate import evaluate, random_input
+from airobas.blocks_hub.mip_blocks_lib.linear.verification_formula.verification_constraints import (
     get_output_constrs,
-    get_output_constrs_specific,
 )
-from verif_pipeline import (
+from airobas.verif_pipeline import (
     BlockVerif,
     BlockVerifOutput,
     DataContainer,

diff --git a/blocks_hub/marabou_block.py → airobas/blocks_hub/marabou_block.py b/blocks_hub/marabou_block.py → airobas/blocks_hub/marabou_block.py
@@ -5,10 +5,10 @@
 import numpy as np
 from maraboupy import Marabou, MarabouCore
 from maraboupy.MarabouNetwork import MarabouNetwork  # (pip install maraboupy)
-from tensorflow.keras.layers import Activation, Dense, Input
+from tensorflow.keras.layers import Activation, Dense
 from tensorflow.keras.models import Sequential
 
-from verif_pipeline import (
+from airobas.verif_pipeline import (
     BlockVerif,
     BlockVerifOutput,
     DataContainer,

diff --git a/blocks_hub/meta_block.py → airobas/blocks_hub/meta_block.py b/blocks_hub/meta_block.py → airobas/blocks_hub/meta_block.py
@@ -2,7 +2,7 @@
 
 import numpy as np
 
-from verif_pipeline import (
+from airobas.verif_pipeline import (
     BlockVerif,
     BlockVerifOutput,
     DataContainer,

diff --git a/blocks_hub/mip_blocks_lib/__init__.py → ...bas/blocks_hub/mip_blocks_lib/__init__.py b/blocks_hub/mip_blocks_lib/__init__.py → ...bas/blocks_hub/mip_blocks_lib/__init__.py
diff --git a/..._blocks_lib/commons/utilities/__init__.py → ...ks_hub/mip_blocks_lib/commons/__init__.py b/..._blocks_lib/commons/utilities/__init__.py → ...ks_hub/mip_blocks_lib/commons/__init__.py
diff --git a/...b/linear/verification_formula/__init__.py → ...ib/commons/bounds_propagation/__init__.py b/...b/linear/verification_formula/__init__.py → ...ib/commons/bounds_propagation/__init__.py
diff --git a/...propagation/bounds_computation_decomon.py → ...propagation/bounds_computation_decomon.py b/...propagation/bounds_computation_decomon.py → ...propagation/bounds_computation_decomon.py
@@ -16,15 +16,15 @@
 
 from tensorflow.python.keras.models import Model  # , Sequential, load_model
 
-from blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_interface import (
-    BoundComputation,
+from airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_interface import (
+    BoundComputation
 )
-from blocks_hub.mip_blocks_lib.commons.bounds_propagation.linear_propagation import (
+from airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.linear_propagation import (
     Bounds,
     compute_bounds_linear_propagation,
 )
-from blocks_hub.mip_blocks_lib.commons.layers import Linear, Relu
-from blocks_hub.mip_blocks_lib.commons.neural_network import (
+from airobas.blocks_hub.mip_blocks_lib.commons.layers import Linear, Relu
+from airobas.blocks_hub.mip_blocks_lib.commons.neural_network import (
     NeuralNetwork,
     neural_network_to_keras,
 )

diff --git a/...opagation/bounds_computation_interface.py → ...opagation/bounds_computation_interface.py b/...opagation/bounds_computation_interface.py → ...opagation/bounds_computation_interface.py
@@ -1,11 +1,10 @@
 from abc import abstractmethod
 
-from blocks_hub.mip_blocks_lib.commons.bounds_propagation.linear_propagation import (
+from airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.linear_propagation import (
     compute_bounds_linear_propagation,
 )
-from blocks_hub.mip_blocks_lib.commons.neural_network import NeuralNetwork
-from blocks_hub.mip_blocks_lib.commons.parameters import Bounds
-
+from airobas.blocks_hub.mip_blocks_lib.commons.neural_network import NeuralNetwork
+from airobas.blocks_hub.mip_blocks_lib.commons.parameters import Bounds
 
 def default_propagation(neural_network: NeuralNetwork):
     compute_bounds_linear_propagation(

diff --git a/...ds_propagation/bounds_computation_milp.py → ...ds_propagation/bounds_computation_milp.py b/...ds_propagation/bounds_computation_milp.py → ...ds_propagation/bounds_computation_milp.py
diff --git a/...bounds_computation_symbolic_arithmetic.py → ...bounds_computation_symbolic_arithmetic.py b/...bounds_computation_symbolic_arithmetic.py → ...bounds_computation_symbolic_arithmetic.py
@@ -1,11 +1,11 @@
-from blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_interface import (
+from airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_interface import (
     BoundComputation,
 )
-from blocks_hub.mip_blocks_lib.commons.bounds_propagation.linear_propagation import (
+from airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.linear_propagation import (
     compute_bounds_linear_propagation,
 )
-from blocks_hub.mip_blocks_lib.commons.neural_network import NeuralNetwork
-from blocks_hub.mip_blocks_lib.commons.parameters import Bounds
+from airobas.blocks_hub.mip_blocks_lib.commons.neural_network import NeuralNetwork
+from airobas.blocks_hub.mip_blocks_lib.commons.parameters import Bounds
 
 
 def default_propagation(neural_network: NeuralNetwork):

diff --git a/...s_propagation/bounds_computation_utils.py → ...s_propagation/bounds_computation_utils.py b/...s_propagation/bounds_computation_utils.py → ...s_propagation/bounds_computation_utils.py
@@ -2,17 +2,13 @@
     from decomon.models.utils import ConvertMethod
 except ImportError as e:
     print(e)
-from blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_decomon import (
+from airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_decomon import (
     BoundsComputationDecomon,
 )
-from blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_interface import (
-    BoundComputation,
-)
-from blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_symbolic_arithmetic import (
+from airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_computation_symbolic_arithmetic import (
     BoundComputationSymbolicArithmetic,
-    default_propagation,
 )
-from blocks_hub.mip_blocks_lib.commons.parameters import (
+from airobas.blocks_hub.mip_blocks_lib.commons.parameters import (
     ParamsBoundComputation,
     ParamsBoundComputationEnum,
 )

diff --git a/...nds_propagation/bounds_function_linear.py → ...nds_propagation/bounds_function_linear.py b/...nds_propagation/bounds_function_linear.py → ...nds_propagation/bounds_function_linear.py
@@ -2,9 +2,9 @@
 
 import numpy as np
 
-from blocks_hub.mip_blocks_lib.commons.layers import Layer
-from blocks_hub.mip_blocks_lib.commons.linearfunctions import LinearFunctions
-from blocks_hub.mip_blocks_lib.commons.parameters import Bounds
+from airobas.blocks_hub.mip_blocks_lib.commons.layers import Layer
+from airobas.blocks_hub.mip_blocks_lib.commons.linearfunctions import LinearFunctions
+from airobas.blocks_hub.mip_blocks_lib.commons.parameters import Bounds
 
 
 def compute_bounds(

diff --git a/...ounds_propagation/bounds_function_relu.py → ...ounds_propagation/bounds_function_relu.py b/...ounds_propagation/bounds_function_relu.py → ...ounds_propagation/bounds_function_relu.py
@@ -2,9 +2,9 @@
 
 import numpy as np
 
-from blocks_hub.mip_blocks_lib.commons.layers import Layer
-from blocks_hub.mip_blocks_lib.commons.linearfunctions import LinearFunctions
-from blocks_hub.mip_blocks_lib.commons.parameters import Bounds
+from airobas.blocks_hub.mip_blocks_lib.commons.layers import Layer
+from airobas.blocks_hub.mip_blocks_lib.commons.linearfunctions import LinearFunctions
+from airobas.blocks_hub.mip_blocks_lib.commons.parameters import Bounds
 
 
 def compute_bounds(

diff --git a/...propagation/bounds_function_relu_numba.py → ...propagation/bounds_function_relu_numba.py b/...propagation/bounds_function_relu_numba.py → ...propagation/bounds_function_relu_numba.py
@@ -3,16 +3,16 @@
 import numpy as np
 from numba import njit
 
-from blocks_hub.mip_blocks_lib.commons.layers import Layer, Relu
-from blocks_hub.mip_blocks_lib.commons.linearfunctions import (
+from airobas.blocks_hub.mip_blocks_lib.commons.layers import Layer, Relu
+from airobas.blocks_hub.mip_blocks_lib.commons.linearfunctions import (
     LinearFunctions,
     compute_max_values_numba,
     compute_min_values_numba,
     compute_upper_numba,
     get_lower_relu_relax_numba,
     get_upper_relu_relax_numba,
 )
-from blocks_hub.mip_blocks_lib.commons.parameters import Bounds
+from airobas.blocks_hub.mip_blocks_lib.commons.parameters import Bounds
 
 
 def compute_bounds_numba(

diff --git a/.../bounds_propagation/linear_propagation.py → .../bounds_propagation/linear_propagation.py b/.../bounds_propagation/linear_propagation.py → .../bounds_propagation/linear_propagation.py
@@ -1,8 +1,8 @@
-import blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_function_linear as bounds_linear
-import blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_function_relu as bounds_relu
-from blocks_hub.mip_blocks_lib.commons.layers import InputLayer, Relu
-from blocks_hub.mip_blocks_lib.commons.neural_network import NeuralNetwork
-from blocks_hub.mip_blocks_lib.commons.parameters import Bounds
+import airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_function_linear as bounds_linear
+import airobas.blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_function_relu as bounds_relu
+from airobas.blocks_hub.mip_blocks_lib.commons.layers import InputLayer, Relu
+from airobas.blocks_hub.mip_blocks_lib.commons.neural_network import NeuralNetwork
+from airobas.blocks_hub.mip_blocks_lib.commons.parameters import Bounds
 
 
 def compute_bounds_linear_propagation(

diff --git a/...s_propagation/linear_propagation_numba.py → ...s_propagation/linear_propagation_numba.py b/...s_propagation/linear_propagation_numba.py → ...s_propagation/linear_propagation_numba.py
@@ -1,7 +1,7 @@
-import blocks_hub.mip_blocks_lib.commons.bounds_propagation.bounds_function_relu_numba as bounds_relu
-from blocks_hub.mip_blocks_lib.commons.layers import InputLayer, Relu
-from blocks_hub.mip_blocks_lib.commons.neural_network import NeuralNetwork
-from blocks_hub.mip_blocks_lib.commons.parameters import Bounds
+from airobas import blocks_hub as bounds_relu
+from airobas.blocks_hub.mip_blocks_lib.commons.layers import InputLayer, Relu
+from airobas.blocks_hub.mip_blocks_lib.commons.neural_network import NeuralNetwork
+from airobas.blocks_hub.mip_blocks_lib.commons.parameters import Bounds
 
 
 def compute_bounds_linear_propagation(

diff --git a/blocks_hub/mip_blocks_lib/commons/formula.py → ...cks_hub/mip_blocks_lib/commons/formula.py b/blocks_hub/mip_blocks_lib/commons/formula.py → ...cks_hub/mip_blocks_lib/commons/formula.py
diff --git a/blocks_hub/mip_blocks_lib/commons/layers.py → ...ocks_hub/mip_blocks_lib/commons/layers.py b/blocks_hub/mip_blocks_lib/commons/layers.py → ...ocks_hub/mip_blocks_lib/commons/layers.py
@@ -5,7 +5,7 @@
 
 import numpy as np
 
-from blocks_hub.mip_blocks_lib.commons.linearfunctions import LinearFunctions
+from airobas.blocks_hub.mip_blocks_lib.commons.linearfunctions import LinearFunctions
 
 
 class LayerType(Enum):

diff --git a/...mip_blocks_lib/commons/linearfunctions.py → ...mip_blocks_lib/commons/linearfunctions.py b/...mip_blocks_lib/commons/linearfunctions.py → ...mip_blocks_lib/commons/linearfunctions.py
diff --git a/.../mip_blocks_lib/commons/neural_network.py → .../mip_blocks_lib/commons/neural_network.py b/.../mip_blocks_lib/commons/neural_network.py → .../mip_blocks_lib/commons/neural_network.py
@@ -11,7 +11,7 @@
 from keras.layers import Activation, Dense
 from keras.models import Sequential
 
-from blocks_hub.mip_blocks_lib.commons.layers import (
+from airobas.blocks_hub.mip_blocks_lib.commons.layers import (
     InputLayer,
     InputType,
     Layer,

diff --git a/..._hub/mip_blocks_lib/commons/parameters.py → ..._hub/mip_blocks_lib/commons/parameters.py b/..._hub/mip_blocks_lib/commons/parameters.py → ..._hub/mip_blocks_lib/commons/parameters.py
diff --git a/airobas/blocks_hub/mip_blocks_lib/commons/utilities/__init__.py b/airobas/blocks_hub/mip_blocks_lib/commons/utilities/__init__.py
diff --git a/..._blocks_lib/commons/utilities/evaluate.py → ..._blocks_lib/commons/utilities/evaluate.py b/..._blocks_lib/commons/utilities/evaluate.py → ..._blocks_lib/commons/utilities/evaluate.py
@@ -1,6 +1,6 @@
 import numpy as np
 
-from blocks_hub.mip_blocks_lib.commons.neural_network import Linear, NeuralNetwork, Relu
+from airobas.blocks_hub.mip_blocks_lib.commons.neural_network import Linear, NeuralNetwork, Relu
 
 
 def random_input(neural_net: NeuralNetwork):

diff --git a/airobas/blocks_hub/mip_blocks_lib/linear/__init__.py b/airobas/blocks_hub/mip_blocks_lib/linear/__init__.py
diff --git a/airobas/blocks_hub/mip_blocks_lib/linear/verification_formula/__init__.py b/airobas/blocks_hub/mip_blocks_lib/linear/verification_formula/__init__.py