An Efficient Explanation of Individual Classifications using Game Theory

@article{trumbelj2010AnEE,
  title={An Efficient Explanation of Individual Classifications using Game Theory},
  author={Erik {\vS}trumbelj and Igor Kononenko},
  journal={J. Mach. Learn. Res.},
  year={2010},
  volume={11},
  pages={1-18}
}
We present a general method for explaining individual predictions of classification models. The method is based on fundamental concepts from coalitional game theory and predictions are explained with contributions of individual feature values. We overcome the method's initial exponential time complexity with a sampling-based approximation. In the experimental part of the paper we use the developed method on models generated by several well-known machine learning algorithms on both synthetic and… Expand
The Explanation Game: Explaining Machine Learning Models with Cooperative Game Theory
TLDR
This work illustrates how subtle differences in the underlying game formulations of existing methods can cause large differences in attribution for a prediction, and proposes a general framework for generating explanations for ML models, called formulate, approximate, and explain (FAE). Expand
The Explanation Game: Explaining Machine Learning Models Using Shapley Values
TLDR
This work illustrates how subtle differences in the underlying game formulations of existing methods can cause large Differences in the attributions for a prediction, and presents a general game formulation that unifies existing methods, and enables straightforward confidence intervals on their attributions. Expand
Explaining prediction models and individual predictions with feature contributions
TLDR
A sensitivity analysis-based method for explaining prediction models that can be applied to any type of classification or regression model, and which is equivalent to commonly used additive model-specific methods when explaining an additive model. Expand
The Shapley Value of coalition of variables provides better explanations
TLDR
A Python library1 that computes reliably conditional expectations and SV for tree-based models, is implemented and compared with state-of-the-art algorithms on toy models and real data sets. Expand
A Game Theoretic Approach to Class-wise Selective Rationalization
TLDR
This work proposes a new game theoretic approach to class-dependent rationalization, where the method is specifically trained to highlight evidence supporting alternative conclusions and is able to identify both factual and counterfactual rationales consistent with human rationalization. Expand
Coalitional Strategies for Efficient Individual Prediction Explanation
TLDR
These coalitional methods based on the detection of relevant groups of attributes influencing a prediction are provided and compares them with the literature and show that they are more efficient than existing ones such as SHapley Additive exPlanation (SHAP). Expand
Explainable Artificial Intelligence: How Subsets of the Training Data Affect a Prediction
TLDR
This paper considers data-driven models which are already developed, implemented and trained and proposes a novel methodology which is called Shapley values for training data subset importance, arguing that the explanations enable us to perceive more of the inner workings of the algorithms, and illustrate how models producing similar predictions can be based on very different parts of the training data. Expand
Explainable Voting
TLDR
It is proved that outcomes of the important Borda rule can be explained using O(m) steps, where m is the number of alternatives, and implies that the foregoing bound is asymptotically tight. Expand
Quality of classification explanations with PRBF
TLDR
The empirically compare the quality of explanations based on marginalization of the Gaussian distribution and explanation with averaging over all feature values and show that the approximation method and the exact solution give very similar results, which increases the confidence in the explanation methodology also for other classification models. Expand
shapr: An R-package for explaining machine learning models with dependence-aware Shapley values
TLDR
A common task within machine learning is to train a model to predict an unknown outcome (response variable) based on a set of known input variables/features, so that they are often viewed as “black-boxes”, that produce some output from some input. Expand
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 29 REFERENCES
Feature Selection via Coalitional Game Theory
TLDR
Empirical comparison with several other existing feature selection methods shows that the backward elimination variant of CSA leads to the most accurate classification results on an array of data sets. Expand
Explaining Classifications For Individual Instances
TLDR
It is demonstrated that the generated explanations closely follow the learned models and a visualization technique is presented that shows the utility of the approach and enables the comparison of different prediction methods. Expand
Polynomial calculation of the Shapley value based on sampling
TLDR
A polynomial method based on sampling theory that can be used to estimate the Shapley value for cooperative games is developed and some desirable statistical properties of the proposed approach are examined. Expand
Explaining instance classifications with interactions of subsets of feature values
TLDR
A novel method for explaining the decisions of an arbitrary classifier, independent of the type of classifiers, that works at the instance level, decomposing the model's prediction for an instance into the contributions of the attributes' values. Expand
Fair Attribution of Functional Contribution in Artificial and Biological Networks
TLDR
The multi-perturbation Shapley value analysis, an axiomatic, scalable, and rigorous method for deducing causal function localization from multiple perturbations data, accurately quantifies the contributions of network elements and their interactions. Expand
Visual Explanation of Evidence with Additive Classifiers
TLDR
A framework, ExplainD, is described for explaining decisions made by classifiers that use additive evidence, which applies to many widely used classifiers, including linear discriminants and many additive models. Expand
Contact personalization using a score understanding method
TLDR
This paper presents a method to interpret the output of a classification (or regression) model based on two concepts: the variable importance and the value importance of the variable. Expand
Wrappers for Feature Subset Selection
TLDR
The wrapper method searches for an optimal feature subset tailored to a particular algorithm and a domain and compares the wrapper approach to induction without feature subset selection and to Relief, a filter approach tofeature subset selection. Expand
Random Forests
  • L. Breiman
  • Mathematics, Computer Science
  • Machine Learning
  • 2004
TLDR
Internal estimates monitor error, strength, and correlation and these are used to show the response to increasing the number of features used in the forest, and are also applicable to regression. Expand
Visualizing the Simple Bayesian Classi er
The simple Bayesian classi er (SBC), sometimes called Naive-Bayes, is built based on a conditional independence model of each attribute given the class. The model was previously shown to beExpand
...
1
2
3
...