aiverify-robustness-toolbox 2.2.0

AI Verify Robustness Toolbox generates a perturbed dataset using boundary attack algorithm on the test dataset.

Algorithm - Robustness Toolbox

License

• Licensed under Apache Software License 2.0

Developers:

• AI Verify

Installation

Each test algorithm can now be installed via pip and run individually.

pip install aiverify-robustness-toolbox

Example Usage:

The robustness plugin supports testing of tabular and image data.

Run the following bash script to execute the plugin on tabular data:

#!/bin/bash

root_path="<PATH_TO_FOLDER>/aiverify/stock-plugins/user_defined_files"
python -m aiverify_robustness_toolbox \
  --data_path $root_path/data/sample_bc_pipeline_credit_data.sav \
  --model_path $root_path/pipeline/bc_tabular_credit \
  --ground_truth_path $root_path/data/sample_bc_pipeline_credit_ytest_data.sav \
  --ground_truth default \
  --model_type CLASSIFICATION \
  --run_pipeline

Run the following bash script to execute the plugin on image data:

#!/bin/bash

root_path="<PATH_TO_FOLDER>/aiverify/stock-plugins/user_defined_files"
python -m aiverify_robustness_toolbox \
  --data_path $root_path/data/raw_fashion_image_10/0.png \
  --model_path $root_path/pipeline/mc_image_fashion \
  --ground_truth_path $root_path/data/pickle_pandas_fashion_mnist_annotated_labels_10.sav \
  --ground_truth label \
  --model_type CLASSIFICATION \
  --run_pipeline \
  --annotated_ground_truth_path $root_path/data/pickle_pandas_fashion_mnist_annotated_labels_10.sav \
  --file_name_label file_name

If the algorithm runs successfully, the results of the test will be saved in an output folder.

Develop plugin locally

Assuming aiverify-test-engine has already been installed in the virtual environment, run the following bash script to install the plugin and execute a test:

#!/bin/bash

# setup virtual environment
python3 -m venv .venv
source .venv/bin/activate

# install plugin
cd aiverify/stock-plugins/aiverify.stock.robustness-toolbox/algorithms/robustness_toolbox
pip install .

python -m aiverify_robustness_toolbox --data_path  <data_path> --model_path <model_path> --ground_truth_path <ground_truth_path> --ground_truth <str> --model_type CLASSIFICATION --run_pipeline --annotated_ground_truth_path <str> --file_name_label <str>

Build Plugin

cd aiverify/stock-plugins/aiverify.stock.robustness-toolbox/algorithms/robustness_toolbox
hatch build

Tests

Pytest is used as the testing framework.

Run the following steps to execute the unit and integration tests inside the tests/ folder

cd aiverify/stock-plugins/aiverify.stock.robustness-toolbox/algorithms/robustness_toolbox
pytest .

Run using Docker

In the aiverify root directory, run the below command to build the docker image

docker build -t aiverify-robustness-toolbox -f stock-plugins/aiverify.stock.robustness-toolbox/algorithms/robustness_toolbox/Dockerfile .

Run the below bash script to run the algorithm

#!/bin/bash
docker run \
  -v $(pwd)/stock-plugins/user_defined_files:/input \
  -v $(pwd)/stock-plugins/aiverify.stock.robustness-toolbox/algorithms/robustness_toolbox/output:/app/aiverify/output \
  aiverify-robustness-toolbox \
  --data_path /input/data/raw_fashion_image_10 \
  --model_path /input/pipeline/mc_image_fashion \
  --ground_truth_path /input/data/pickle_pandas_fashion_mnist_annotated_labels_10.sav \
  --ground_truth label \
  --model_type CLASSIFICATION \
  --run_pipeline \
  --annotated_ground_truth_path /input/data/pickle_pandas_fashion_mnist_annotated_labels_10.sav \
  --file_name_label file_name

If the algorithm runs successfully, the results of the test will be saved in an output folder in the algorithm directory.

Tests

Pytest is used as the testing framework.

Run the following steps to execute the unit and integration tests inside the tests/ folder

docker run \
  --entrypoint python3 \
  -w /app/aiverify/stock-plugins/aiverify.stock.robustness-toolbox/algorithms/robustness_toolbox \
  aiverify-robustness-toolbox \
  -m pytest .