ART 1.13.1

[beat-buesser]

This release of ART 1.13.1 provides updates to ART 1.13

Added

[None]

Changed

• Changed PDTP privacy metric to support two comparison: ratio (default) and new difference mode (Small enhancement to PDTP privacy metric #1984)
• Changed default parameters for apply_fit and apply_predict for the Data Augmentation defenses CutMix*, CutOut*, and MixUp* (DP-InstaHide and Data Augmentation Defenses Bugfix #1987)

Removed

[None]

Fixed

• Fixed bug in PixelThreshold attack to support batches of a single sample (Fix bug in PixelThreshold for single input #1982)
• Fixed type error in DPInstaHideTrainer for PyTorchClassifier by casting random noise to correct type (DP-InstaHide and Data Augmentation Defenses Bugfix #1987)
• Added missing classes to union types OBJECT_DETECTOR_TYPE, PYTORCH_ESTIMATOR_TYPE, and TENSORFLOWV2_ESTIMATOR_TYPE (Add Missing Object Detection Estimator Types #1999)
• Fixed audio perturbations going out of clip values in insert_tone_trigger and insert_audio_trigger (Clip Audio Perturbations in Range #2016)
• Fixed missing transfer to device in FeatureAdversariesPyTorch to enable running on GPUs (Add missing transfer to device in FeatureAdversariesPyTorch #2021)
• Fixed missing covnersion to float to support floor() on GPUs in PyTorchClassifier (Add conversion to float for comparison on GPU/CUDA #2022)
• Fixed incorrect integer return type in check_and_transform_label_format (Cast nb_classes to integer in check_and_transform_label_format #2025)

---

This discussion was created from the release ART 1.13.1.