MS-CPFI: A model-agnostic Counterfactual Perturbation Feature Importance algorithm for interpreting black-box Multi-State models.
Aziliz CottinMarine ZulianNicolas PécuchetAgathe GuillouxSandrine KatsahianPublished in: Artif. Intell. Medicine (2024)
Keyphrases
- probabilistic model
- black box
- multiple models
- objective function
- parametric models
- learned models
- bayesian framework
- learning algorithm
- state space
- parameter estimation
- high level
- similarity measure
- databases
- expectation maximization
- feature importance
- optimization method
- bayesian networks
- information retrieval
- machine learning