Considerations when learning additive explanations for black-box models
From MaRDI portal
Publication:6176233
Abstract: Many methods to explain black-box models, whether local or global, are additive. In this paper, we study global additive explanations for non-additive models, focusing on four explanation methods: partial dependence, Shapley explanations adapted to a global setting, distilled additive explanations, and gradient-based explanations. We show that different explanation methods characterize non-additive components in a black-box model's prediction function in different ways. We use the concepts of main and total effects to anchor additive explanations, and quantitatively evaluate additive and non-additive explanations. Even though distilled explanations are generally the most accurate additive explanations, non-additive explanations such as tree explanations that explicitly model non-additive components tend to be even more accurate. Despite this, our user study showed that machine learning practitioners were better able to leverage additive explanations for various tasks. These considerations should be taken into account when considering which explanation to trust and use to explain black-box models.
Recommendations
- Explaining individual predictions when features are dependent: more accurate approximations to Shapley values
- A comparison of instance-level counterfactual explanation algorithms for behavioral and textual data: SEDC, LIME-C and SHAP-C
- How to explain individual classification decisions
- A survey on the explainability of supervised machine learning
- Interpreting deep learning models with marginal attribution by conditioning on quantiles
Cites work
- All models are wrong, but many are useful: learning a variable's importance by studying an entire class of prediction models simultaneously
- Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models
- Generalized additive models
- Generalized additive models. An introduction with R.
- Greedy function approximation: A gradient boosting machine.
- Interpretable classifiers using rules and Bayesian analysis: building a better stroke prediction model
- Learning certifiably optimal rule lists for categorical data
- On concurvity in nonlinear and nonparametric regression models
- On sensitivity estimation for nonlinear mathematical models
- Predictive learning via rule ensembles
- Prototype selection for interpretable classification
- Random forests
- Sobol' indices and Shapley value
- Visualizing the Effects of Predictor Variables in Black Box Supervised Learning Models
Cited in
(7)- A comparison of instance-level counterfactual explanation algorithms for behavioral and textual data: SEDC, LIME-C and SHAP-C
- Local and global explanations of agent behavior: integrating strategy summaries with saliency maps
- Unrestricted permutation forces extrapolation: variable importance requires at least one more model, or there is no free variable importance
- Non-monotonic explanation functions
- Learning a functional control for high-frequency finance
- Explainable generalized additive neural networks with independent neural network training
- SLISEMAP: supervised dimensionality reduction through local explanations
This page was built for publication: Considerations when learning additive explanations for black-box models
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6176233)
