Improvement-Focused Causal Recourse (ICR)
نویسندگان
چکیده
Algorithmic recourse recommendations inform stakeholders of how to act revert unfavorable decisions. However, existing methods may recommend actions that lead acceptance (i.e., the model's decision) but do not improvement underlying real-world state). To such is fooling predictor. We introduce a novel method, Improvement-Focused Causal Recourse (ICR), which involves conceptual shift: Firstly, we require ICR guide toward improvement. Secondly, tailor be accepted by specific Instead, leverage causal knowledge design decision systems predict accurately pre- and post-recourse, guarantees translate into guarantees. Curiously, optimal pre-recourse classifiers are robust thus suitable post-recourse. In semi-synthetic experiments, demonstrate given correct ICR, in contrast approaches, guides both
منابع مشابه
Causal Networks Based Process Improvement
This paper includes a causal-based modelling of software process models in order to analyse the correct relationships between the different (key) process areas of these models. A first short description of causal network approaches shows the identified problems and possible benefits using these formal techniques in the software engineering area. The definition and extension of the causal modell...
متن کاملModification of Recourse Data for Mixed-Integer Recourse Models
We consider modification of the recourse data, consisting of the second-stage parameters and the underlying distribution, as an approximation technique for solving two-stage recourse problems. This approach is applied to several specific classes of mixed-integer recourse problems; in each case, the resulting recourse problem is much easier to solve.
متن کاملSimplification of recourse models by modification of recourse data
We consider modification of the recourse data, consisting of the second-stage parameters and the underlying distribution, as an approximation technique for solving two-stage recourse problems. This approach is applied to several specific classes of recourse problems; in each case, the resulting recourse problem is much easier to solve. Modification of recourse data is shown to be the common pri...
متن کاملProcess Improvement Focused Analysis of VINST IT Support Logs
The BLOCKIN BLOCKIN goal BLOCKIN
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the ... AAAI Conference on Artificial Intelligence
سال: 2023
ISSN: ['2159-5399', '2374-3468']
DOI: https://doi.org/10.1609/aaai.v37i10.26398