Marco Anisetti

Marco Anisetti, Claudio A. Ardagna, Alessandro Balestrucci et al.

Machine learning is becoming ubiquitous. From finance to medicine, machine learning models are boosting decision-making processes and even outperforming humans in some tasks. This huge progress in terms of prediction quality does not however find a counterpart in the security of such models and corresponding predictions, where perturbations of fractions of the training set (poisoning) can seriously undermine the model accuracy. Research on poisoning attacks and defenses received increasing attention in the last decade, leading to several promising solutions aiming to increase the robustness of machine learning. Among them, ensemble-based defenses, where different models are trained on portions of the training set and their predictions are then aggregated, provide strong theoretical guarantees at the price of a linear overhead. Surprisingly, ensemble-based defenses, which do not pose any restrictions on the base model, have not been applied to increase the robustness of random forest models. The work in this paper aims to fill in this gap by designing and implementing a novel hash-based ensemble approach that protects random forest against untargeted, random poisoning attacks. An extensive experimental evaluation measures the performance of our approach against a variety of attacks, as well as its sustainability in terms of resource consumption and performance, and compares it with a traditional monolithic model based on random forest. A final discussion presents our main findings and compares our approach with existing poisoning defenses targeting random forests.

Marco Anisetti, Claudio A. Ardagna, Nicola Bena et al.

Modern applications are increasingly driven by Machine Learning (ML) models whose non-deterministic behavior is affecting the entire application life cycle from design to operation. The pervasive adoption of ML is urgently calling for approaches that guarantee a stable non-functional behavior of ML-based applications over time and across model changes. To this aim, non-functional properties of ML models, such as privacy, confidentiality, fairness, and explainability, must be monitored, verified, and maintained. Existing approaches mostly focus on i) implementing solutions for classifier selection according to the functional behavior of ML models, ii) finding new algorithmic solutions, such as continuous re-training. In this paper, we propose a multi-model approach that aims to guarantee a stable non-functional behavior of ML-based applications. An architectural and methodological approach is provided to compare multiple ML models showing similar non-functional properties and select the model supporting stable non-functional behavior over time according to (dynamic and unpredictable) contextual changes. Our approach goes beyond the state of the art by providing a solution that continuously guarantees a stable non-functional behavior of ML-based applications, is ML algorithm-agnostic, and is driven by non-functional properties assessed on the ML models themselves. It consists of a two-step process working during application operation, where model assessment verifies non-functional properties of ML models trained and selected at development time, and model substitution guarantees continuous and stable support of non-functional properties. We experimentally evaluate our solution in a real-world scenario focusing on non-functional property fairness.

Panagiotis Katrakazas, Aristotelis Ballas, Marco Anisetti et al.

Colorectal cancer is the third most common tumor in men and the second in women, accounting for 10% of all tumors worldwide. It ranks second in cancer-related deaths with 9.4%, following lung cancer. The decrease in mortality rate documented over the last 20 years has shown signs of slowing down since 2017, necessitating concentrated actions on specific measures that have exhibited considerable potential. As such, the technical foundation and research evidence for blood-derived protein markers have been set, pending comparative validation, clinical implementation and integration into an artificial intelligence enabled decision support framework that also considers knowledge on risk factors. The current paper aspires to constitute the driving force for creating change in colorectal cancer screening by reviewing existing medical practices through accessible and non-invasive risk estimation, employing a straightforward artificial intelligence outlook.

Marco Anisetti, Claudio A. Ardagna, Nicola Bena et al.

Machine Learning (ML) is increasingly used to implement advanced applications with non-deterministic behavior, which operate on the cloud-edge continuum. The pervasive adoption of ML is urgently calling for assurance solutions assessing applications non-functional properties (e.g., fairness, robustness, privacy) with the aim to improve their trustworthiness. Certification has been clearly identified by policymakers, regulators, and industrial stakeholders as the preferred assurance technique to address this pressing need. Unfortunately, existing certification schemes are not immediately applicable to non-deterministic applications built on ML models. This article analyzes the challenges and deficiencies of current certification schemes, discusses open research issues, and proposes a first certification scheme for ML-based applications.

Marco Anisetti, Claudio A. Ardagna, Michele Bezzi et al.

Privacy-aware processing of personal data on the web of services requires managing a number of issues arising both from the technical and the legal domain. Several approaches have been proposed to matching privacy requirements (on the clients side) and privacy guarantees (on the service provider side). Still, the assurance of effective data protection (when possible) relies on substantial human effort and exposes organizations to significant (non-)compliance risks. In this paper we put forward the idea that a privacy certification scheme producing and managing machine-readable artifacts in the form of privacy certificates can play an important role towards the solution of this problem. Digital privacy certificates represent the reasons why a privacy property holds for a service and describe the privacy measures supporting it. Also, privacy certificates can be used to automatically select services whose certificates match the client policies (privacy requirements). Our proposal relies on an evolution of the conceptual model developed in the Assert4Soa project and on a certificate format specifically tailored to represent privacy properties. To validate our approach, we present a worked-out instance showing how privacy property Retention-based unlinkability can be certified for a banking financial service.