adversarial-robustness-toolbox

adversarial-robustness-toolbox copied to clipboard

Support for multioutput model

6

Is it possible to train a multi-output model similar to the following [example](https://github.com/Trusted-AI/adversarial-robustness-toolbox/blob/main/examples/get_started_tensorflow_v2.py)? I tried with a custom `train_step` function and multi output model similar to the one in example...

aqibsaeed

**Is your feature request related to a problem? Please describe.** Remove all of the duplicated and slightly different fixture `def fix_get_mnist_subset(get_mnist_dataset):` from each test and unify them to all using...

killianlevacher

**Is your feature request related to a problem? Please describe.** Now that we are using github actions and are no longer limited by the Travis time limit, I would suggest...

killianlevacher

"Only Keras classifiers (v2.2.4) are supported for this defence" in poisoning_defense_neural_cleanse.ipynd

5

Dear all, how to guarantee I am using v2.2.4 when I run this defense test? I try to install keras v2.2.4, but it does not work. Many thanks.

authwork

detection_adversarial_samples_cifar10.ipynb is not working correctly on nbviewer. The reason is tf version.

2

It's easy to reproduce this error through the tutorial of nbviewer, and the problem is the tf version on the server is lower than expected 2.2.0(I found the same error...

YacratesWyh

Poison Detection - Activation Defense with Keras Generator

3

**Describe the bug** If a batch of a generator doesn't contain all classes the method segmenation of classes returns an empty list for the class and the line 671 cannot...

e117777

Currently, we don't have a default set of audio samples for testing purposes available. The Lingvo ASR estimator uses three Librispeech samples, which it downloads from the Qin et al....

hesseltuinhof

Implement a model-specific estimator for TensorFlow SSD object detection models (https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf1_detection_zoo.md). The implementation can likely follow a similar pattern as for `TensorFlowFasterRCNN` estimator and we should investigate if multiple/all base...

beat-buesser

Enable PGD attack on PyTorch Faster-RCNN using np object arrays as input

1

**Is your feature request related to a problem? Please describe.** I'm looking to run PGD on ART's PyTorch Faster-RCNN using the xView dataset. This dataset contains images of varying shapes,...

lcadalzo

**Is your feature request related to a problem? Please describe.** Add a method for abstaining classifiers that returns something different than NP array. This would make it clear for end...

ebubae