Recent Publications
Personalized Algorithmic Recourse with Preference Elicitation
Giovanni De Toni, Paolo Viappiani, Stefano Teso, Bruno Lepri, and Andrea Passerini.
In
Transactions on Machine Learning Research, 2024.
@article {tmlr2024,
author = { Toni, Giovanni De and Viappiani, Paolo and Teso, Stefano and Lepri, Bruno and Passerini, Andrea },
title = "Personalized Algorithmic Recourse with Preference Elicitation",
journal = "Transactions on Machine Learning Research",
issn = "2835-8856",
year = "2024",
url = "https://openreview.net/forum?id=8sg2I9zXgO",
note = "",
}
Enhancing SMT-based Weighted Model Integration by structure awareness
Giuseppe Spallitta, Gabriele Masina, Paolo Morettin, Andrea Passerini, and Roberto Sebastiani.
In
Artificial Intelligence, 2024.
[paper]
[bibtex]
[abstract]
The development of efficient exact and approximate algorithms for probabilistic inference is a long-standing goal of artificial intelligence research. Whereas substantial progress has been made in dealing with purely discrete or purely continuous domains, adapting the developed solutions to tackle hybrid domains, characterized by discrete and continuous variables and their relationships, is highly non-trivial. Weighted Model Integration (WMI) recently emerged as a unifying formalism for probabilistic inference in hybrid domains. Despite a considerable amount of recent work, allowing WMI algorithms to scale with the complexity of the hybrid problem is still a challenge. In this paper we highlight some substantial limitations of existing state-of-the-art solutions, and develop an algorithm that combines SMT-based enumeration, an efficient technique in formal verification, with an effective encoding of the problem structure. This allows our algorithm to avoid generating redundant models, resulting in drastic computational savings. Additionally, we show how SMT-based approaches can seamlessly deal with different integration techniques, both exact and approximate, significantly expanding the set of problems that can be tackled by WMI technology. An extensive experimental evaluation on both synthetic and real-world datasets confirms the substantial advantage of the proposed solution over existing alternatives. The application potential of this technology is further showcased on a prototypical task aimed at verifying the fairness of probabilistic programs.
@article {aij2024,
author = { Spallitta, Giuseppe and Masina, Gabriele and Morettin, Paolo and Passerini, Andrea and Sebastiani, Roberto },
title = "Enhancing SMT-based Weighted Model Integration by structure awareness",
journal = "Artificial Intelligence",
volume = "328",
pages = "104067",
year = "2024",
issn = "0004-3702",
doi = "https://doi.org/10.1016/j.artint.2024.104067",
url = "https://www.sciencedirect.com/science/article/pii/S0004370224000031",
keywords = "Hybrid probabilistic inference, Weighted Model Integration, Satisfiability modulo theories",
}
Generating fine-grained surrogate temporal networks
Antonio Longa, Giulia Cencetti, Sune Lehmann, Andrea Passerini, and Bruno Lepri.
In
Communications Physics 7(22), 2024.
@article {commphys2024,
author = { Longa, Antonio and Cencetti, Giulia and Lehmann, Sune and Passerini, Andrea and Lepri, Bruno },
title = "Generating fine-grained surrogate temporal networks",
year = "2024",
journal = "Communications Physics",
volume = "7",
number = "22",
url = "https://arxiv.org/abs/2205.08820",
}
Environmentally-Aware Bundle Recommendation Using the Choquet Integral
Marco Bronzini, Erich Robbi, Paolo Viappiani, and Andrea Passerini.
In
{12th Conference on Prestigious Applications of Intelligent Systems (PAIS 2023)} (Frontiers in Artificial Intelligence and Applications), 2023.
@inproceedings {pais2023choquet,
author = { Bronzini, Marco and Robbi, Erich and Viappiani, Paolo and Passerini, Andrea },
title = "{Environmentally-Aware Bundle Recommendation Using the Choquet Integral}",
booktitle = "{12th Conference on Prestigious Applications of Intelligent Systems (PAIS 2023)}",
address = "Krakow, Poland, Poland",
publisher = "{IOS Press}",
series = "Frontiers in Artificial Intelligence and Applications",
number = "372",
pages = "3182--3189",
year = "2023",
month = "September",
doi = "10.3233/FAIA230639",
url = "https://hal.science/hal-04292392/file/FAIA-372-FAIA230639.pdf",
}
Synthesizing explainable counterfactual policies for algorithmic recourse with program synthesis
Giovanni De Toni, Bruno Lepri, and Andrea Passerini.
In
Mach. Learn. 112(4), 2023.
[paper]
[bibtex]
[abstract]
Being able to provide counterfactual interventions—sequences of actions we would have had to take for a desirable outcome to happen—is essential to explain how to change an unfavourable decision by a black-box machine learning model (e.g., being denied a loan request). Existing solutions have mainly focused on generating feasible interventions without providing explanations of their rationale. Moreover, they need to solve a separate optimization problem for each user. In this paper, we take a different approach and learn a program that outputs a sequence of explainable counterfactual actions given a user description and a causal graph. We leverage program synthesis techniques, reinforcement learning coupled with Monte Carlo Tree Search for efficient exploration, and rule learning to extract explanations for each recommended action. An experimental evaluation on synthetic and real-world datasets shows how our approach, FARE (eFficient counterfActual REcourse), generates effective interventions by making orders of magnitude fewer queries to the black-box classifier with respect to existing solutions, with the additional benefit of complementing them with interpretable explanations.
@article {mach2023,
author = { De Toni, Giovanni and Lepri, Bruno and Passerini, Andrea },
title = "Synthesizing explainable counterfactual policies for algorithmic recourse with program synthesis",
year = "2023",
issue_date = "Apr 2023",
publisher = "Kluwer Academic Publishers",
address = "USA",
volume = "112",
number = "4",
issn = "0885-6125",
url = "https://doi.org/10.1007/s10994-022-06293-7",
doi = "10.1007/s10994-022-06293-7",
journal = "Mach. Learn.",
month = "feb",
pages = "1389–1409",
numpages = "21",
keywords = "Machine learning, Explainable AI, Counterfactuals examples, Algorithmic recourse",
}
Semantic Loss Functions for Neuro-Symbolic Structured Prediction
Kareem Ahmed, Stefano Teso, Paolo Morettin, Luca Di Liello, Pierfrancesco Ardino, Jacopo Gobbi, Yitao Liang, Eric Wang, Kai{-}Wei Chang, Andrea Passerini, and Guy Van Broeck.
In
Unknown venue (type=incollection), 2023.
@incollection {faia2023,
author = { Ahmed, Kareem and Teso, Stefano and Morettin, Paolo and Liello, Luca Di and Ardino, Pierfrancesco and Gobbi, Jacopo and Liang, Yitao and Wang, Eric and Chang, Kai{-}Wei and Passerini, Andrea and den Broeck, Guy Van },
title = "Semantic Loss Functions for Neuro-Symbolic Structured Prediction",
booktitle = "Compendium of Neurosymbolic Artificial Intelligence",
series = "Frontiers in Artificial Intelligence and Applications",
volume = "369",
pages = "485--505",
publisher = "{IOS} Press",
year = "2023",
url = "https://doi.org/10.3233/FAIA230154",
doi = "10.3233/FAIA230154",
}
Adaptation of Student Behavioural Routines during COVID-19: A Multimodal Approach
Nicolò A. Girardini, Simone Centellegher, Andrea Passerini, Ivano Bison, Fausto Giunchiglia, and Bruno Lepri.
In
EPJ Data Science 12(55), 2023.
@article {epjdatascience2023,
author = { Girardini, Nicolò A. and Centellegher, Simone and Passerini, Andrea and Bison, Ivano and Giunchiglia, Fausto and Lepri, Bruno },
title = "Adaptation of Student Behavioural Routines during COVID-19: A Multimodal Approach",
year = "2023",
journal = "EPJ Data Science",
volume = "12",
number = "55",
url = "papers/epjdatascience2023.pdf",
}
Interpretability Is in the Mind of the Beholder: A Causal Framework for Human-Interpretable Representation Learning
Emanuele Marconato, Andrea Passerini, and Stefano Teso.
In
Entropy 25(12), 2023.
@article {entropy2023,
author = { Marconato, Emanuele and Passerini, Andrea and Teso, Stefano },
title = "Interpretability Is in the Mind of the Beholder: A Causal Framework for Human-Interpretable Representation Learning",
journal = "Entropy",
volume = "25",
year = "2023",
number = "12",
article-number = "1574",
url = "https://www.mdpi.com/1099-4300/25/12/1574",
doi = "10.3390/e25121574",
}
See complete publication list