Adversarial Robustness Toolbox v1.5.2 releases: crafting and analysis of attacks and defense methods for machine learning models

Adversarial Robustness Toolbox

Adversarial Robustness 360 Toolbox (ART) is a Python library supporting developers and researchers in defending Machine Learning models (Deep Neural Networks, Gradient Boosted Decision Trees, Support Vector Machines, Random Forests, Logistic Regression, Gaussian Processes, Decision Trees, Scikit-learn Pipelines, etc.) against adversarial threats and helps to make AI systems more secure and trustworthy. Machine Learning models are vulnerable to adversarial examples, which are inputs (images, texts, tabular data, etc.) deliberately modified to produce a desired response by the Machine Learning model. ART provides the tools to build and deploy defenses and test them with adversarial attacks.

Defending Machine Learning models involves certifying and verifying model robustness and model hardening with approaches such as pre-processing inputs, augmenting training data with adversarial samples, and leveraging runtime detection methods to flag any inputs that might have been modified by an adversary. The attacks implemented in ART allow creating adversarial attacks against Machine Learning models which are required to test defenses with state-of-the-art threat models.

Supported attack and defense methods

The Adversarial Robustness Toolbox contains implementations of the following attacks:

• Deep Fool (Moosavi-Dezfooli et al., 2015)
• Fast Gradient Method (Goodfellow et al., 2014)
• Jacobian Saliency Map (Papernot et al., 2016)
• Universal Perturbation (Moosavi-Dezfooli et al., 2016)
• Virtual Adversarial Method (Moosavi-Dezfooli et al., 2015)
• C&W Attack (Carlini and Wagner, 2016)
• NewtonFool (Jang et al., 2017)

The following defense methods are also supported:

• Feature squeezing (Xu et al., 2017)
• Spatial smoothing (Xu et al., 2017)
• Label smoothing (Warde-Farley and Goodfellow, 2016)
• Adversarial training (Szegedy et al., 2013)
• Virtual adversarial training (Miyato et al., 2017)

The details of the work from IBM research can be found in the research paper. The ART toolbox is developed with the goal of helping developers better understand

• Measuring model robustness
• Model hardening
• Runtime detection

Changelog v1.5.2

Added

• Added new method reset_patch to art.attacks.evasion.adversarial_patch.* to reset patch (#863)
• Added passing kwargs to internal attacks of art.attacks.evasion.AutoAttack (#850)
• Added art.estimators.classification.BlackBoxClassifierNeuralNetwork as black-box classifier for neural network models (#849)
• Added support for channels_first=False for art.attacks.evasion.ShadowAttack in PyTorch (#848)

Changed

• Changed Numpy requirements to be less strict to resolve conflicts in dependencies (#879)
• Changed estimator requirements for art.attacks.evasion.SquareAttack and art.attacks.evasion.SimBA to include NeuralNetworkMixin requiring neural network models (#849)

Fixed

• Fixed BaseEstimator.set_params to set preprocessing and preprocessing_defences correctly by accounting for art.preprocessing.standardisation_mean_std (#901)
• Fixed support for CUDA in art.attacks.inference.membership_inference.MembershipInferenceBlackBox.infer (#899)
• Fixed return in art.preprocessing.standardisation_mean_std.StandardisationMeanStdPyTorch to maintain correct dtype (#890)
• Fixed type conversion in art.evaluations.security_curve.SecurityCurve to be explicit (#886)
• Fixed dtype in art.attacks.evasion.SquareAttack for norm=2 to maintain correct type (#877)
• Fixed missing CarliniWagnerASR in art.attacks.evasion namespace (#873)
• Fixed support for CUDA i `art.estimators.classification.PyTorchClassifier.loss (#862)
• Fixed bug in art.attacks.evasion.AutoProjectedGradientDescent for targeted attack to correctly detect successful iteration steps and added robust stopping criteria if loss becomes zero (#860)
• Fixed bug in initialisation of search space in art.attacks.evasion.SaliencyMapMethod (#843)
• Fixed bug in support for video data in art.attacks.evasion.adversarial_patch.AdversarialPatchNumpy (#838)
• Fixed bug in logged success rate of art.attacks.evasion.ProjectedGradientDescentPyTorch and art.attacks.evasion.ProjectedGradientDescentTensorFlowV2 to use correct labels (#833)

Download && Tutorial

Copyright (C) IBM Corporation 2018