The term counterfactual

For more info about our services contact : help@bestpfe.com

Table of contents

1 Introduction 
2 Technical Context 
2.1 Key Notions of Machine Learning Interpretability
2.2 Surrogate Model Approaches
2.3 Counterfactual Explanation Approaches
2.4 Conclusion
2.5 Notations
3 Generating Post-hoc Counterfactuals and the Risk of Out-of-distribution Explanations 
3.1 Motivations
3.2 Proposed Problem Formalization and the Growing Spheres Algorithm .
3.3 Experimental Validation
3.4 Discussion: Out-of-Distribution Counterfactuals
3.5 Conclusion
4 The Risk of Unjustified Explanations 
4.1 Ground-truth Justification
4.2 LRA: an Algorithm to Detect Unjustified Classification Regions
4.3 Experimental Assessment of the Local Risk of Generating Unjustified Counterfactuals
4.4 VE: An Algorithm to Assess the Vulnerability of Post-hoc Counterfactual
Approaches
4.5 Conclusion
5 Defining Explanation Locality for Post-hoc Surrogate Models 
5.1 Locality for Local Surrogate Models
5.2 Measuring Locality: the Local Fidelity Criterion
5.3 A New Local Surrogate Approach: the LS Algorithm
5.4 Discussion: Local Surrogates and Counterfactuals
5.5 Conclusion
6 Conclusion and Perspectives 
6.1 Summary of the Contributions
6.2 FutureWorks
Appendix A Justification of Adversarial Examples on MNIST 
References