Grokking Anchors: Uncovering What a Machine-Learning Model Relies On

Kilian Kluge @ PyConDE & PyData Berlin 2023, April 19th 2023, 11:50, A1

Links & further reading

• Ribeiro et al. (2018): Anchors: High-Precision Model-Agnostic Explanations
• Goerke & Lang: Scoped Rules (Anchors) (part of Christopher Molnar: Interpretable Machine Learning)
• Kaufmann,& Kalyanakrishnan (2013): Information Complexity in Bandit Subset Selection
• Kluge & Eckhardt (2020): Explaining Suspected Phishing Attempts with Document Anchors 2020 ICML Workshop on Human Interpretability in Machine Learning (WHI 2020)
• Anchor implementations:
  • Original Anchors on GitHub & package on PyPI
  • Documentation for Anchors in alibi & package on PyPI
• Affiliations:
  • inlinity (LinkedIn)
  • XAI Studio (LinkedIn)