151. Counterfactual explanations under uncertainty
Invited abstract in session MC-12: Robust optimisation and its applications, stream Applications: AI, uncertainty management and sustainability.
Monday, 14:00-16:00Room: B100/8009
Authors (first author is the speaker)
| 1. | Antonio Navas-Orozco
|
| Departamento de Estadística e Investigación Operativa, Universidad de Sevilla | |
| 2. | Emilio CARRIZOSA
|
| IMUS - Instituto de Matemáticas de la Universidad de Sevilla |
Abstract
In a Generalized Linear Model (GLM), finding a counterfactual decision of a record amounts to finding a vector with maximal outcome at a distance small enough of the record.
In this paper, we extend this problem and address the challenge of building robust counterfactual decisions. Robustness is understood as guaranteeing that the predicted outcome for the counterfactual decision remains sufficiently high when the nominal probability distribution (the empirical distribution for the given training sample) is replaced by a probability distribution with the same support but different frequencies.
Exploiting the structure of the Exponential Family, the problem of finding a robust counterfactual decision is expressed as a biobjective bilevel nonlinear optimization problem, whose structural properties are studied, and a numerical method is proposed.
Keywords
- Optimization for learning and data analysis
- Data driven optimization
- Distributionally robust optimization
Status: accepted
Back to the list of papers