On the Art and Science of Machine Learning Explanations

Abstract

This text discusses several popular explanatory methods that go beyond the error measurements and plots traditionally used to assess machine learning models. Some of the explanatory methods are accepted tools of the trade while others are rigorously derived and backed by long-standing theory. The methods, decision tree surrogate models, individual conditional expectation (ICE) plots, local interpretable model-agnostic explanations (LIME), partial dependence plots, and Shapley explanations, vary in terms of scope, fidelity, and suitable application domain. Along with descriptions of these methods, this text presents real-world usage recommendations supported by a use case and public, in-depth software examples for reproducibility.

arXiv commentThis manuscript is a preprint of the text for an invited talk at the 2019 KDD XAI workshop. A previous version has also appeared in the proceedings of the Joint Statistical Meetings. Errata and updates available here: https://github.com/jphall663/kdd_2019. Version 2 incorporated reviewer feedback. Version 3 includes a minor adjustment to Figure 1. Version 4 corrects a minor typo

References in corpus (13)

• Towards A Rigorous Science of Interpretable Machine Learning
• Axiomatic Attribution for Deep Networks
• Learning Important Features Through Propagating Activation Differences
• Physics Informed Deep Learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equations
• Consistent Individualized Feature Attribution for Tree Ensembles
• Generating Natural Adversarial Examples
• Supersparse Linear Integer Models for Optimized Medical Scoring Systems
• Consistent feature attribution for tree ensembles
• Interpreting Blackbox Models via Model Extraction
• Explainable Neural Networks based on Additive Index Models
• Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead
• Locally Interpretable Models and Effects based on Supervised Partitioning (LIME-SUP)
• Fairwashing: the risk of rationalization

Cited by in corpus (7)

• Explainable Matrix -- Visualization for Global and Local Interpretability of Random Forest Classification Ensembles
• Landscape of R packages for eXplainable Artificial Intelligence
• Proposed Guidelines for the Responsible Use of Explainable Machine Learning
• Wasserstein-based fairness interpretability framework for machine learning models
• SAFE ML: Surrogate Assisted Feature Extraction for Model Learning
• Sampling - Variational Auto Encoder - Ensemble: In the Quest of Explainable Artificial Intelligence
• Lifting Interpretability-Performance Trade-off via Automated Feature Engineering