[wiki work in progress]

ART 1.9 introduces a Summary Writer API which collects metrics about losses, gradients, adversarial patches and textures, etc. at strategic locations in the supported attacks. ART 1.9 provides a default implementation art.summary_writer.SummaryWriterDefault and allows users to implement custom summary writers by implementing base class art.summary_writer.SummaryWriter.

• art.attacks.evasion.ProjectedGradientDescent*
• art.attacks.evasion.AdversarialPatch*
• art.attacks.evasion.AdversrialTexture*

The attack’s Summary Writer is activated with argument summary_writer:

• Boolean: Default writer and output in current location
• String: Default writer and output in defined path
• SummaryWriter: Custom writer in custom location

• All metrics as function of iteration step

• Silent Success (I₁)

Tracks if adversarial example switches back to benign.

• Break-point Angle (I₂)

Quantifies attack convergence.

• Increasing Loss (I₃)

Quantifies attack convergence rate.

• Zero Gradients (I₄)

Checks quality of loss gradients.

Users can create custom summary writers to collect custom metrics by implementing art.summary_writer.SummaryWriter and adding custom metrics in method update:

def update(self, batch_id, global_step, grad=None, patch=None, estimator=None, x=None, y=None, **kwargs):

providing access to current batch ID and global step, gradients, patches, estimator, input and label data.