Research Interests and Scientific Activity
Mariagrazia Fugini’s research interests are in Security and in Information Systems as follows:
- Data Security and Privacy
- Information Systems Development
- Public Administration Information Systems
- Risk Mitigation in Work Environments
- Safety of Cyber-Physical Systems
- Green Information Systems
- AI and Healthcare: Interactions with AI systems
In particular, some themes are:
1. Data Security
- access control to data;
- acquisition and specification of security policies and requirements;
- design for security;
- security in advanced systems: web-based information systems, cloud computing, workflows, e-commerce, e-services, power grids.
Ph.D. Thesis on security management in databases (in cooperation with University of Maryland, College Park). Cooperation with research teams and companies.
1.1 Research on Intrusion Detection Systems and Fraud Detection (from 2024)
- Adaptive Real-Time Financial Fraud Detection with Explainable AI tools
Financial crime is increasingly facilitated by technology and globalization, demanding advanced IT tools for detection. The Fraud Detection System proposed in this research is devised to operate under pragmatic operational constraints inherent to financial institutions, such as extreme class imbalance, due to the rarity of fraudulent events, or the continuously evolving fraud patterns (concept drift) driven by adversarial adaptation, and, most significant, the stochastic delays in obtaining verified feedback, crucial for model supervision.
This research develops financial transaction monitoring in a data stream context; it details a developed streaming Machine Learning (ML) pipeline, designed with a lightweight yet powerful Data Stream Management System (DSMS) for real-time feature engineering, and a multi-stage analytical engine that orchestrates diverse detection logic. This engine integrates deterministic rules with adaptive ML models and is coupled with a dynamic decision threshold management system to optimize the precision-recall trade-off under operational pressures. A core contribution is the systematic, empirical comparison of diverse adaptive learning strategies, ranging from instance-incremental to various batch-incremental methods, to assess their adaptability and effectiveness under these operational conditions. Furthermore, the research aims at an actionable interpretability framework designed to synthesize low-level feature attributions into user-centric concept importances, enhancing the utilities at the disposal of investigators.
Ref Paper: https://dl.acm.org/doi/pdf/10.1145/3794859
- Dynamic Multi-Objective Ensemble Control for Cost-Aware Intrusion Detection Systems
- Network intrusion detection is a critical component of network
security monitoring, and it is essential to detect intrusions in real
time. This paper proposes a resource-aware orchestration framework that
redefines Network Intrusion Detection as a control problem, addressing
the trade-off between high-fidelity deep learning architectures and the latency
constraints of high-throughput streams. Our approach integrates
heterogeneous classifiers within a weighted soft-voting ensemble architecture
designed to provide detection performance while managing computational
load. Existing ensemble methods typically aggregate all models
at all times, incurring high computational cost regardless of threat complexity.
To avoid this, our framework promotes or prunes models using a
real-time utility function based on accuracy, diversity, and latency. Experimental
evaluation on diverse datasets indicates that non-linear models
contribute to detection of complex threat vectors, while lightweight
linear filters demonstrate adequate performance for attacks where aggressive
streaming learners exhibit catastrophic forgetting. Results quantify
these trade-offs, showing a reduction in the variance of adaptive components
across the evaluated datasets, and providing an approach for
designing AI components within security infrastructures.
Ref paper: FRAUD-ENSEMBLE-LEARNING DEXA26
2. Information Systems Development
- specification of distributed, web based applications;
- design tools development;
- re-engineering of information systems;
- distributed cooperative workflows, business processes and e-services;
- information systems for Public Administrations;
- portals for services to citizens and companies, E-Government.
Main EU and National Projects:
- ITHACA – Integrated Toolkit for Highly Advanced Commercial Applications, 1989- 1993
- F3 – From Formal to Fuzzy, 1992-1994
- WIDE (Workflow Interactive Development Environment) 1990-1992
- DEAFIN (Project for harmonization of information systems in European Public Administrations) 1998-2000
- WS-Diamond (Diagnosis and Monitoring of Self-Healing Web Service Applications) 2005-2008
- S-cube (Service Network of Excellence – currently active) 2008-2012
- TOISE (Trusted Computing for European Embedded devices), 2011- 2013
Cooperation with BarcelonaTech (UPC), UNIMI-Bicocca, UNIMI, Center for Information Technology Renato Archer – CTI, Brazil, University of Seville, University of Toulouse.
International reviewer of PhD Thesis and Master Students exchange – Anna University, Chennai, India; Stockholm University,Sweden; University of Toulouse, France and other.
3. Public Administration Information Systems
Research on services for PAs in support to employment, healthcare, social care for frail people. She co-developed the SINTESI and BLL systems.
Main EU and National Projects:
- SEEMP (Single European Employment MarketPlace 2005-2009
- Various Projects of the Italian Ministry of Research and of CNR
- Borsa Lavoro Lombardia, SINTESI, Borsa Nazionale del Lavoro, 2002 -2007
- Attiv@bili (2014-2016)
- Big Data Analytics (2016-2018)
4. Risk Mitigation in Work Environments
Development of software platforms for risk management out of data from mobile sensors. Risk mitigation strategies, analysis of business convenience of ensuring assets. Techniques for personal protection. Development of prototypes for risk analysis (with probabilistic techniques) and emergency management (with deterministic techniques).
Cooperation with Frederick University Nicosia, Cyprus, UNIMI-Bicocca, Fondazione Politecnico di Milano and companies in the “Smart Cities” Tender (Lineacom, Microdevice, MailUp).
5. Co-Engineering Security and Safety of Cyber-Physical Systems
Methodologies and tools for facing security threats and safety hazards that generate risks in smart environment s(e.g., work areas). Methods to capture and model risk events using inference and knowledge representation techniques in a “World of Things”.
Cooperation with Department of Energy Politecnico di Milano, King Saud University, Dept. of Computer Science, Università degli Studi di Milano (SESAR Group), Department of Electrical and Computer Engineering,Khalifa University, Abu Dabi.
6. Green Information Systems
Main EU and National Projects:
- GAMES (2010-2012)
- ECO2Clouds (2012-2014)
Research Activity Summary
Co-author of “Database Security”, Addison Wesley, 1995, “Sicurezza dei Sistemi Informatici”, Apogeo 2001, “Sistemi Informativi Basati su Web”, F.Angeli 2006, “Impianti Informatici”, McGraw-Hill”, 2007 and of various publications on information system design and management, data security, service-based applications, business process management and risk management in work environments. PC members of various Conferences and Journals in Security and Information Systems. Program Chair of WETICE 2017.
7. AI and Healthcare: Interactions with AI systems
Research on AI systems for healthcare, in particular, medical interaction with explainable AI.