Defuse: Harnessing Unrestricted Adversarial Examples for Debugging Models Beyond Test Accuracy

no code implementations • 11 Feb 2021 • Dylan Slack, Nathalie Rauschmayr, Krishnaram Kenthapadi

Each region contains a specific type of model bug; for instance, a misclassification region for an MNIST classifier contains a style of skinny 6 that the model mistakes as a 1.

BIG-bench Machine Learning

Defuse: Debugging Classifiers Through Distilling Unrestricted Adversarial Examples

no code implementations • 1 Jan 2021 • Dylan Z Slack, Nathalie Rauschmayr, Krishnaram Kenthapadi

As a route to better discover and fix model bugs, we propose failure scenarios: regions on the data manifold that are incorrectly classified by a model.

Clustering