Explaining the world, one model at a time.


I'm Mattia Setzu, a research fellow at the Department of Computer Science of the University of Pisa, Italy. I focus on Explainable AI, with the broad research goal of automating the understanding of complex machine learning models.

  1. Rule Relevance Score

    RRS (Rule-Relevance Score) is a scoring-based approach tackling the Local to Global problem. …


  2. The Local to Global problem

    Machine learning finds application in a wide range of areas, some of which are sensitive to discrimination. Upon inspection, classifiers have repeatedly shown to learn biases on sex, race, and other sensitive attributes 1. Explainable AI techniques addr...…


  3. Research at Aarhus University

    In the past summer I have spent two months at Aarhus University to collaborate with the Data Intensive System Group led by Prof. Ira Assent. For the two months of my visit, I collaborated with Simon Enni. Together we have developed an agnostic methodolo...…


  4. Master's thesis and the Local to Global problem

    The local 2 global problem is a data-starved generalization problem. Data-starved problems aim to generalize local models to the larger input space through the use of oracles and adversarial algorithms. …