Dino Pedreschi

ECAI Conference 2025 Conference Paper

Group Explainability Through Local Approximation

• Mattia Setzu
• Riccardo Guidotti
• Dino Pedreschi
• Fosca Giannotti

Machine learning models are becoming increasingly complex and widely adopted. Interpretable machine learning allows us to not only make predictions but also understand the rationale behind automated decisions through explanations. Explanations are typically characterized by their scope: local explanations are generated by local surrogate models for specific instances, while global explanations aim to approximate the behavior of the entire black-box model. In this paper, we break this dichotomy of locality to explore an underexamined area that lies between these two extremes: meso-level explanations. The goal of meso-level explainability is to provide explanations using a set of meso-level interpretable models, which capture patterns at an intermediate level of abstraction. To this end, we propose GROUX, an explainable-by-design algorithm that generates meso-level explanations in the form of feature importance scores. Our approach includes a partitioning phase that identifies meso groups, followed by the training of interpretable models within each group. We evaluate GROUX on a collection of tabular datasets, reporting both the accuracy and complexity of the resulting meso models, and compare it against other meso-level explainability algorithms. Additionally, we analyze the algorithm’s sensitivity to its hyperparameters to better understand its behavior and robustness.

AIJ Journal 2025 Journal Article

Human-AI coevolution

• Dino Pedreschi
• Luca Pappalardo
• Emanuele Ferragina
• Ricardo Baeza-Yates
• Albert-László Barabási
• Frank Dignum
• Virginia Dignum
• Tina Eliassi-Rad

Human-AI coevolution, defined as a process in which humans and AI algorithms continuously influence each other, increasingly characterises our society, but is understudied in artificial intelligence and complexity science literature. Recommender systems and assistants play a prominent role in human-AI coevolution, as they permeate many facets of daily life and influence human choices through online platforms. The interaction between users and AI results in a potentially endless feedback loop, wherein users' choices generate data to train AI models, which, in turn, shape subsequent user preferences. This human-AI feedback loop has peculiar characteristics compared to traditional human-machine interaction and gives rise to complex and often “unintended” systemic outcomes. This paper introduces human-AI coevolution as the cornerstone for a new field of study at the intersection between AI and complexity science focused on the theoretical, empirical, and mathematical investigation of the human-AI feedback loop. In doing so, we: (i) outline the pros and cons of existing methodologies and highlight shortcomings and potential ways for capturing feedback loop mechanisms; (ii) propose a reflection at the intersection between complexity science, AI and society; (iii) provide real-world examples for different human-AI ecosystems; and (iv) illustrate challenges to the creation of such a field of study, conceptualising them at increasing levels of abstraction, i.e., scientific, legal and socio-political.

IJCAI Conference 2025 Conference Paper

Human-AI Coevolution (Abstract Reprint)

• Dino Pedreschi
• Luca Pappalardo
• Emanuele Ferragina
• Ricardo Baeza-Yates
• Albert-László Barabási
• Frank Dignum
• Virginia Dignum
• Tina Eliassi-Rad

Human-AI coevolution, defined as a process in which humans and AI algorithms continuously influence each other, increasingly characterises our society, but is understudied in artificial intelligence and complexity science literature. Recommender systems and assistants play a prominent role in human-AI coevolution, as they permeate many facets of daily life and influence human choices through online platforms. The interaction between users and AI results in a potentially endless feedback loop, wherein users' choices generate data to train AI models, which, in turn, shape subsequent user preferences. This human-AI feedback loop has peculiar characteristics compared to traditional human-machine interaction and gives rise to complex and often “unintended” systemic outcomes. This paper introduces human-AI coevolution as the cornerstone for a new field of study at the intersection between AI and complexity science focused on the theoretical, empirical, and mathematical investigation of the human-AI feedback loop. In doing so, we: (i) outline the pros and cons of existing methodologies and highlight shortcomings and potential ways for capturing feedback loop mechanisms; (ii) propose a reflection at the intersection between complexity science, AI and society; (iii) provide real-world examples for different human-AI ecosystems; and (iv) illustrate challenges to the creation of such a field of study, conceptualising them at increasing levels of abstraction, i. e. , scientific, legal and socio-political.

AIJ Journal 2021 Journal Article

GLocalX - From Local to Global Explanations of Black Box AI Models

• Mattia Setzu
• Riccardo Guidotti
• Anna Monreale
• Franco Turini
• Dino Pedreschi
• Fosca Giannotti

Artificial Intelligence (AI) has come to prominence as one of the major components of our society, with applications in most aspects of our lives. In this field, complex and highly nonlinear machine learning models such as ensemble models, deep neural networks, and Support Vector Machines have consistently shown remarkable accuracy in solving complex tasks. Although accurate, AI models often are “black boxes” which we are not able to understand. Relying on these models has a multifaceted impact and raises significant concerns about their transparency. Applications in sensitive and critical domains are a strong motivational factor in trying to understand the behavior of black boxes. We propose to address this issue by providing an interpretable layer on top of black box models by aggregating “local” explanations. We present GLocalX, a “local-first” model agnostic explanation method. Starting from local explanations expressed in form of local decision rules, GLocalX iteratively generalizes them into global explanations by hierarchically aggregating them. Our goal is to learn accurate yet simple interpretable models to emulate the given black box, and, if possible, replace it entirely. We validate GLocalX in a set of experiments in standard and constrained settings with limited or no access to either data or local explanations. Experiments show that GLocalX is able to accurately emulate several models with simple and small models, reaching state-of-the-art performance against natively global solutions. Our findings show how it is often possible to achieve a high level of both accuracy and comprehensibility of classification models, even in complex domains with high-dimensional data, without necessarily trading one property for the other. This is a key requirement for a trustworthy AI, necessary for adoption in high-stakes decision making applications.

AAAI Conference 2020 Conference Paper

Explaining Image Classifiers Generating Exemplars and Counter-Exemplars from Latent Representations

• Riccardo Guidotti
• Anna Monreale
• Stan Matwin
• Dino Pedreschi

We present an approach to explain the decisions of black box image classifiers through synthetic exemplar and counterexemplar learnt in the latent feature space. Our explanation method exploits the latent representations learned through an adversarial autoencoder for generating a synthetic neighborhood of the image for which an explanation is required. A decision tree is trained on a set of images represented in the latent space, and its decision rules are used to generate exemplar images showing how the original image can be modified to stay within its class. Counterfactual rules are used to generate counter-exemplars showing how the original image can “morph” into another class. The explanation also comprehends a saliency map highlighting the areas that contribute to its classification, and areas that push it into another class. A wide and deep experimental evaluation proves that the proposed method outperforms existing explainers in terms of fidelity, relevance, coherence, and stability, besides providing the most useful and interpretable explanations.

IS Journal 2019 Journal Article

Factual and Counterfactual Explanations for Black Box Decision Making

• Riccardo Guidotti
• Anna Monreale
• Fosca Giannotti
• Dino Pedreschi
• Salvatore Ruggieri
• Franco Turini

The rise of sophisticated machine learning models has brought accurate but obscure decision systems, which hide their logic, thus undermining transparency, trust, and the adoption of artificial intelligence (AI) in socially sensitive and safety-critical contexts. We introduce a local rule-based explanation method, providing faithful explanations of the decision made by a black box classifier on a specific instance. The proposed method first learns an interpretable, local classifier on a synthetic neighborhood of the instance under investigation, generated by a genetic algorithm. Then, it derives from the interpretable classifier an explanation consisting of a decision rule, explaining the factual reasons of the decision, and a set of counterfactuals, suggesting the changes in the instance features that would lead to a different outcome. Experimental results show that the proposed method outperforms existing approaches in terms of the quality of the explanations and of the accuracy in mimicking the black box.

AAAI Conference 2019 Conference Paper

Meaningful Explanations of Black Box AI Decision Systems

• Dino Pedreschi
• Fosca Giannotti
• Riccardo Guidotti
• Anna Monreale
• Salvatore Ruggieri
• Franco Turini

Black box AI systems for automated decision making, often based on machine learning over (big) data, map a user’s features into a class or a score without exposing the reasons why. This is problematic not only for lack of transparency, but also for possible biases inherited by the algorithms from human prejudices and collection artifacts hidden in the training data, which may lead to unfair or wrong decisions. We focus on the urgent open challenge of how to construct meaningful explanations of opaque AI/ML systems, introducing the local-toglobal framework for black box explanation, articulated along three lines: (i) the language for expressing explanations in terms of logic rules, with statistical and causal interpretation; (ii) the inference of local explanations for revealing the decision rationale for a specific case, by auditing the black box in the vicinity of the target instance; (iii), the bottom-up generalization of many local explanations into simple global ones, with algorithms that optimize for quality and comprehensibility. We argue that the local-first approach opens the door to a wide variety of alternative solutions along different dimensions: a variety of data sources (relational, text, images, etc.), a variety of learning problems (multi-label classification, regression, scoring, ranking), a variety of languages for expressing meaningful explanations, a variety of means to audit a black box.

TIST Journal 2019 Journal Article

PlayeRank

• Luca Pappalardo
• Paolo Cintia
• Paolo Ferragina
• Emanuele Massucco
• Dino Pedreschi
• Fosca Giannotti

The problem of evaluating the performance of soccer players is attracting the interest of many companies and the scientific community, thanks to the availability of massive data capturing all the events generated during a match (e.g., tackles, passes, shots, etc.). Unfortunately, there is no consolidated and widely accepted metric for measuring performance quality in all of its facets. In this article, we design and implement PlayeRank, a data-driven framework that offers a principled multi-dimensional and role-aware evaluation of the performance of soccer players. We build our framework by deploying a massive dataset of soccer-logs and consisting of millions of match events pertaining to four seasons of 18 prominent soccer competitions. By comparing PlayeRank to known algorithms for performance evaluation in soccer, and by exploiting a dataset of players’ evaluations made by professional soccer scouts, we show that PlayeRank significantly outperforms the competitors. We also explore the ratings produced by PlayeRank and discover interesting patterns about the nature of excellent performances and what distinguishes the top players from the others. At the end, we explore some applications of PlayeRank—i.e. searching players and player versatility—showing its flexibility and efficiency, which makes it worth to be used in the design of a scalable platform for soccer analytics.

IS Journal 2017 Journal Article

The Inductive Constraint Programming Loop

• Christian Bessiere
• Luc De Raedt
• Tias Guns
• Lars Kotthoff
• Mirco Nanni
• Siegfried Nijssen
• Barry O'Sullivan
• Anastasia Paparrizou

Constraint programming is used for a variety of real-world optimization problems, such as planning, scheduling, and resource allocation problems, all while we continuously gather vast amounts of data about these problems. Current constraint programming software doesn't exploit such data to update schedules, resources, and plans. The authors propose a new framework that they call the inductive constraint programming loop. In this approach, data is gathered and analyzed systematically to dynamically revise and adapt constraints and optimization criteria. Inductive constraint programming aims to bridge the gap between the areas of data mining and machine learning on one hand and constraint programming on the other.

TIST Journal 2016 Journal Article

Driving Profiles Computation and Monitoring for Car Insurance CRM

• Mirco Nanni
• Roberto Trasarti
• Anna Monreale
• Valerio Grossi
• Dino Pedreschi

Customer segmentation is one of the most traditional and valued tasks in customer relationship management (CRM). In this article, we explore the problem in the context of the car insurance industry, where the mobility behavior of customers plays a key role: Different mobility needs, driving habits, and skills imply also different requirements (level of coverage provided by the insurance) and risks (of accidents). In the present work, we describe a methodology to extract several indicators describing the driving profile of customers, and we provide a clustering-oriented instantiation of the segmentation problem based on such indicators. Then, we consider the availability of a continuous flow of fresh mobility data sent by the circulating vehicles, aiming at keeping our segments constantly up to date. We tackle a major scalability issue that emerges in this context when the number of customers is large—namely, the communication bottleneck—by proposing and implementing a sophisticated distributed monitoring solution that reduces communications between vehicles and company servers to the essential. We validate the framework on a large database of real mobility data coming from GPS devices on private cars. Finally, we analyze the privacy risks that the proposed approach might involve for the users, providing and evaluating a countermeasure based on data perturbation.

AILAW Journal 2014 Journal Article

Anonymity preserving sequential pattern mining

• Anna Monreale
• Dino Pedreschi
• Ruggero G. Pensa
• Fabio Pinelli

Abstract The increasing availability of personal data of a sequential nature, such as time-stamped transaction or location data, enables increasingly sophisticated sequential pattern mining techniques. However, privacy is at risk if it is possible to reconstruct the identity of individuals from sequential data. Therefore, it is important to develop privacy-preserving techniques that support publishing of really anonymous data, without altering the analysis results significantly. In this paper we propose to apply the Privacy-by-design paradigm for designing a technological framework to counter the threats of undesirable, unlawful effects of privacy violation on sequence data, without obstructing the knowledge discovery opportunities of data mining technologies. First, we introduce a k -anonymity framework for sequence data, by defining the sequence linking attack model and its associated countermeasure, a k -anonymity notion for sequence datasets, which provides a formal protection against the attack. Second, we instantiate this framework and provide a specific method for constructing the k -anonymous version of a sequence dataset, which preserves the results of sequential pattern mining, together with several basic statistics and other analytical properties of the original data, including the clustering structure. A comprehensive experimental study on realistic datasets of process-logs, web-logs and GPS tracks is carried out, which empirically shows how, in our proposed method, the protection of privacy meets analytical utility.

AILAW Journal 2010 Journal Article

Integrating induction and deduction for finding evidence of discrimination

• Salvatore Ruggieri
• Dino Pedreschi
• Franco Turini

Abstract We present a reference model for finding (prima facie) evidence of discrimination in datasets of historical decision records in socially sensitive tasks, including access to credit, mortgage, insurance, labor market and other benefits. We formalize the process of direct and indirect discrimination discovery in a rule-based framework, by modelling protected-by-law groups, such as minorities or disadvantaged segments, and contexts where discrimination occurs. Classification rules, extracted from the historical records, allow for unveiling contexts of unlawful discrimination, where the degree of burden over protected-by-law groups is evaluated by formalizing existing norms and regulations in terms of quantitative measures. The measures are defined as functions of the contingency table of a classification rule, and their statistical significance is assessed, relying on a large body of statistical inference methods for proportions. Key legal concepts and reasonings are then used to drive the analysis on the set of classification rules, with the aim of discovering patterns of discrimination, either direct or indirect. Analyses of affirmative action, favoritism and argumentation against discrimination allegations are also modelled in the proposed framework. Finally, we present an implementation, called LP2DD, of the overall reference model that integrates induction, through data mining classification rule extraction, and deduction, through a computational logic implementation of the analytical tools. The LP2DD system is put at work on the analysis of a dataset of credit decision records.

CSL Conference 1999 Conference Paper

On the Effective Semantics of Nondeterministic, Nonmonotonic, Temporal Logic Databases

• Fosca Giannotti
• Giuseppe Manco 0001
• Mirco Nanni
• Dino Pedreschi

Abstract We consider in this paper an extension of Datalog with mechanisms for temporal, nonmonotonic and nondeterministic reasoning, which we refer to as Datalog++. We study its semantics, and show how iterated fixpoint and stable model semantics can be combined to the purpose of clarifying the interpretation of Datalog++ programs, and supporting their efficient execution. On this basis, the design of appropriate optimization techniques for Datalog++ is also discussed.

ECAI Conference 1990 Conference Paper

Universal Quantification by Case Analysis

• Antonio Brogi
• Paolo Mancarella
• Dino Pedreschi
• Franco Turini