Conference talk, JuliaCon Local Paris 2025, Paris, France
Symbolic learning is a branch of machine learning that studies algorithms for building symbolic models, classifiers that can be translated to logical rules. Hence, differently from neural networks and other statistical models, symbolic models are more easily readable and interpretable. Common examples include decision trees and their ensemble counterpart, random forests.
Poster session, 36th European Summer School in Logic, Language and Information, Ruhr University Bochum
Reasoning tasks over modal logics are notoriously difficult, comprising problems which are often semi-decidable. Hence, it is not rare for applications leveraging such tasks to exploit heuristic approaches able to give results in an acceptable amount of time, at the cost of some inaccuracies. Some approaches may benefit from vector embeddings, allowing for instance to involve neural networks and machine learning models in the reasoning process. This work aims at providing a new way to provide such embeddings specifically for the modal case, allowing for faster reasoning through both mathematical and learning techniques.
Conference talk, Logic, Algebra, and Truth Degrees (LATD 2025), University of Siena, Siena, Italy
Modal logics offer a valid treatment for temporal and spatial data, which are critical in modeling many real-world scenarios and, therefore, are becoming more popular by the day in artificial intelligence applications, specifically when dealing with symbolic machine learning. Some notable examples are [1], [2], introducing modal logics for treating interval temporal relations and topological (i.e., spatial) relations, respectively. However, practitioners handling temporal and spatial data typically encounter challenges, as sensing and discretizing signals that often introduce inaccuracies in the data. Fuzzy logics are renowned as a common approach to deal with uncertainty and unclear boundaries in the data. Furthermore, Melvin Fitting proposed in [3] a many-valued approach leveraging Heyting algebras to tackle many-expert scenarios, another compelling application in artificial intelligence. In this talk, we want to present a framework that is general enough to treat modal many-valued logics, including Fitting’s proposal, and can be endowed with reasoning tools suitable for real-world applications.
Conference talk, Australasian Association for Logic Conference 2024, University of Sydney, Sydney, New South Wales, Australia
Workshop talk, Brisbane Logic Workshop, University of Queensland, Brisbane, Queensland, Australia
Taking inspiration from the literature fuzzy decision trees, and leveraging many-valued logics, we propose a novel, and more general variety of decision trees.