Lectures

You can download the lectures here. We will try to upload lectures prior to their corresponding classes.

• (xml-0) Introduction
  tl;dr: Course logistics and motivation toards Explainability in Machine Learning.
  [slides]
  

  Suggested Readings:

  • Ch-3.1 from Interpretable Machine Learning book by Christoph Molnar
  • Towards A Rigorous Science of Interpretable Machine Learning, Doshi-Velez and Kim, 2017
  • Interpretable machine learning: definitions, methods, and applications, Murdoch et al. 2019
  • Explanation in artificial intelligence: Insights from the social sciences, Tim Miller, 2019
  • Examples are not Enough, Learn to Criticize! Criticism for Interpretability, Kim et al. 2016
• (xml-1) Taxonomy, Scope, and Evaluation of Explainability
  tl;dr: Attending different notions and properties surrounding Explainability.
  [slides]
  

  Suggested Readings:

  • Ch 3.2 to 3.6 from Interpretable Machine Learning book by Christoph Molnar
• (xml-2) Inherently Interpretable Models (Linear and Logistic Regression)
  tl;dr: Interpretability in the basic linear models
  [slides]
  

  Suggested Readings:

  • Ch 3.1, 4.1 to 4.3 from Introduction to Statistical Learning
  • Ch 3.2 to 3.6 from Interpretable Machine Learning book by Christoph Molnar
• (xml-3) Inherently Interpretable Models (GLM, GLA, DT)
  tl;dr: Interpretability in the generalized linear models and Decision Trees
  [slides]
  

  Suggested Readings:

  • Ch 5.3 to 5.7 from Interpretable Machine Learning book by Christoph Molnar
• (xml-4) Model-Agnostic Methods - PDP and ALE
  tl;dr: Global methods describe the average behavior of a machine learning model; Partial Dependence Plots and Accumulated Local Effects
  [slides]
  

  Suggested Readings:

  • Ch 6, Ch 7, Ch 8.1 and 8.2 from Interpretable Machine Learning book by Christoph Molnar
• (xml-5) Model-Agnostic Methods - feature interactions, surrogate, etc.
  tl;dr: Global methods describe the average behavior of a machine learning model - some more methods in the global category
  [slides]
  

  Suggested Readings:

  • Ch 8.3 and 8.7 from Interpretable Machine Learning book by Christoph Molnar
• (xml-6) Local Model-Agnostic Methods - LIME.
  tl;dr: Locally approximating a black-box ML model with an interpretable surrogate
  [slides]
  

  Suggested Readings:

  • LIME, SIGKDD, 2016
• (xml-7) Local Model-Agnostic Methods - Counterfactual Explanations.
  tl;dr: Imagining a hypothetical reality to explain predictions of individual instances
  [slides]
  

  Suggested Readings:

  • Counterfactual Explanations without Opening the Black Box (Harvard Journal of Law & Technology, 2018)
  • Multi-Objective Counterfactual Explanations (PPSN, 2020)
  • Counterfactuals and Causability in XAI
• (xml-8) Interpreting Neural Networks (Feature Visualization)
  tl;dr: Methods to understand the units (learned) in DNNs
  [slides]
  

  Suggested Readings:

  • Distil on Visualization
• (xml-9) Interpreting Neural Networks (Pixel Attribution)
  tl;dr: Highlighting the input features (e.g., pixels) that are relevant for a DNNprediction
  [slides]
  

  Suggested Readings:
• (xml-10) Interpreting Neural Networks (Concept-based Explanations)
  tl;dr: Learning a low-dim representation that can faithfully explain the downstream task of the DNN
  [slides]
  

  Suggested Readings:
• (xml-11) Causal Inference and Explainability
  tl;dr: Linking Causalily (cause and effect) and Explainability in Machine learning
  [slides]
  

  Suggested Readings:

  • Primer on Causal Inference by Judea pearl et al.
  • The Book of Why by Judea Pearl and Dana Mackenzie
  • Handbook on Causality
  • Causality101.net
  • Survey on XAI and Causality by Gianluca Carloni et al., September 2023
  • Surevy on XMl and Causality by Moraffah et al. KDD 2020
• (xml-12) Attention and Explanations
  tl;dr: Does the Attention mechanism provide Explanation for the model?
  [slides]
  

  Suggested Readings:

• (xml-13) Generating Robust Counterfactuals (Algorithmic Recourse)
  tl;dr: Guest lecture: What factors need to be considered while generating counterfactuals for achieving algorithmic recourse?
  [slides]
  

  Suggested Readings:

• (xml-14) Causal Antehoc Explanations
  tl;dr: Guest lecture: Causal associations in neural networks (direct, indirect, and total causal effects)
  [slides]
  

  Suggested Readings:

• (xml-15) Mechanistic Interpretability
  tl;dr: Guest lecture: On the internals of LLMs
  [slides]
  

  Suggested Readings: