Topics covered

Isabelle's lecture

• Intro
  • Explainability and interpretability -- what is it (on a high level) and why do we need it?
  • Types of explainability
• Part 1: Decision understanding
  • Instance-level explainability for text classification and fact checking
    • Language generation based explanations (Atanasova et al., ACL 2020)
    • Evaluating instance-level explanations (Atanasova et al., EMNLP 2020 long)
• Part 2: Model understanding
  • Model-wide explainability for text classification and fact checking
    • Finding model-wide explanations (Atanasova et al. EMNLP 2020 short)
    • Visualising model-wide explanations (Rethmeier et al., UAI 2020)
• Wrap-up
  • Take-aways
  • Future work
  • What have I not covered / pointers to other resources

Niels' lab exercises

• Neuron change probing with TX-Ray

Pepa's lab exercises

• Post-hoc explainability measures and diagnostic properties