Tipo di tesi
Tesi di laurea magistrale
Titolo
Counterfactual Reasoning and Learning in Causal Abstraction
Corso di studi
DATA SCIENCE AND BUSINESS INFORMATICS
Riassunto (Italiano)
A Causal Abstraction establishes a relationship between two Structural Causal Mod-
els (SCMs ) at two different levels of granularity, ensuring that each high-level inter-
ventional distribution is consistent with an intervention performed on the low-level
model. In the present work we focus on linear SCMs and linear abstractions. The
present work tackles the problem of expressing and solving counterfactuals of aggre-
gate variables, within the framework of Causal Abstraction. Furthermore, the work
proposes a methodology to fit a set of linear abstraction functions and a high-level
SCM given a low-level SCM. The proposed methodology ensures the consistency on
low-level and high-level SCMs of a given counterfactual query.
