Shapley-PC: Constraint-based Causal Structure Learning with Shapley Values
F. Russo and F. Toni, 2023. "Shapley-PC: Constraint-based Causal Structure Learning with Shapley Values." Technical Report. arxiv.
Research
My research interests revolve around augumenting human decision making with AI. This involves evaluating the interaction of the desired outcome with its context. Current practices in data-driven decisioning adopt models that span from regression to neural networks and ensemble trees which, although powerful and mostly informative, only capture correlations in the data. Correlation is hardly enough when trying to understand the effect of actions. What decision-makers need is causality.
My PhD project, supervised by Prof. Francesca Toni at Imperial College, works towards creating transparent techniques for causal discovery and its use in machine learning predictions, leveraging argumentation. Causal discovery is about drawing a graph of the relationships influencing an observed phenomenon. Argumentation is a form of non-monotonic reasoning useful for resolving coflicts in a debate. We are creating a system that allows humans and machines to debate about causality. This will highlight cause-effect relationships, giving decision-makers the levers to drive the outcomes they seek.
Talks & Publications
Causal Discovery and Knowledge Injection for Contestable Neural Networks
Russo, Fabrizio and Francesca Toni, 2023. "Causal Discovery and Knowledge Injection for Contestable Neural Networks." in Proceedings of the 26th European Conference on Artificial Intelligence ECAI 2023, Krakow, September, 2023
Argumentation for Interactive Causal Discovery
Fabrizio Russo, 2023. "Argumentation for Interactive Causal Discovery" , in Proceedings of the 32nd International Joint Conference on Artificial Intelligence. Macao, China, August, 2023.
Explaining Classifiers’ Outputs with Causal Models and Argumentation
A. Rago, F. Russo, E. Albini, P. Baroni, F. Toni, 2023. "Explaining Classifiers’ Outputs with Causal Models and Argumentation." Journal of Applied Logics — IfCoLog Journal of Logics and their Applications, Special Issue: Advances in Argumentgation in AI, Volume 10, Number 3: May, 2023.
(Talk) Argumentative Causal Discovery
Fabrizio Russo, "Argumentative Causal Discovery" , STAI Student Seminar Series, Imperial College London, UK.
Causal Discovery and Injection for Feed-Forward Neural Networks
F. Russo and F. Toni, 2022. "Causal Discovery and Injection for Feed-Forward Neural Networks." Technical Report. arxiv.
(Talk) From Credit Risk to Explainable AI Research
F. Russo, "From Credit Risk to Explainable AI Research." Guest Lecture for Master in Finthech - Politecnico di Milano at UCL, London, UK.
Forging Argumentative Explanations from Causal Models
A. Rago, F. Russo, E. Albini, P. Baroni, F. Toni, 2021. "Forging Argumentative Explanations from Causal Models." Proceedings of the 5th Workshop on Advances in Argumentation in Artificial Intelligence 2021 co-located with the 20th International Conference of the Italian Association for Artificial Intelligence (AIxIA 2021). Milan, Italy, November 29th, 2021.
(Talk) Causal Injection into Neural Networks
F. Russo and F. Toni, 2021. "Causal Injection into Neural Networks." 1st International Workshop on Explainable AI in Finance (XAI-FIN) part of the 2021 ACM International Conference on AI in Finance (ICAIF).
(Talk) Risk Scorecards with Machine Learning
F. Russo, T. Ringsjø, D. Smith, J. Woodcock, T. Pile, L. Koteva, "Risk Scorecards with Machine Learning." Modelling with big data and machine learning: interpretability and model uncertainty, Joint Conference by the Bank of England and the Data Analytics for Finance and Macro Research Centre at King’s College London, 2019.
(Talk) Credit Risk Modelling: Data and Techniques Used in the UK Banking Industry
F. Russo, "Credit Risk Modelling: Data and Techniques Used in the UK Banking Industry." THE USE OF CREDIT REGISTER DATA FOR FINANCIAL STABILITY PURPOSES AND CREDIT RISK ANALYSIS, Danmarks Nationalbank Conference, 2019.