ART 1.9.1

[beat-buesser]

This release of ART 1.9.1 provides updates to ART 1.9.

Added

• Added support for TensorFlow 1.15 as backend in KerasClassifier.compute_loss. (Add support in KerasClassifier.compute_loss for TF 1.15 as backend #1466)
• Added support for input range [0, 1] in art.defences.preprocessor.VideoCompression*. (Preprocessing defences in PyTorchGoturn and VideoCompression for [0, 1] range #1470)

Changed

[None]

Removed

[None]

Fixed

• Fixed bug in art.utils.load_nursery for loading nursery dataset with argument raw=True. (Fix bug when loading nursery data  #1460)
• Fixed import of matplotlib to keep it an optional dependency. (Fix import of matplotlib in laser attack utils #1467)
• Fixed bug to allow preprocessing defences to be applied in PyTorchGoturn.predict by adding back missing sample dimension. (Preprocessing defences in PyTorchGoturn and VideoCompression for [0, 1] range #1470)
• Fixed bug in PyTorchClassifier.get_activations to also apply preprocessing if argument framework=True. This fix likely changes the results obtained with BullseyePolytopeAttackPyTorch, the main attack using framework=True. (Replace x with x_preprocessed in PyTorchClassifier.get_activations #1471)

---

This discussion was created from the release ART 1.9.1.