Bruno Casella is a postdoctoral researcher at the Computer Science Department of the University of Turin. His research focuses on federated, distributed, and theoretically grounded learning methods.
He received his PhD in Modeling and Data Science at UniTO on the 26.06.2025 with a thesis entitled “Advancing Federated Learning. Towards Decentralization and Personalized Models“.
He graduated in Computer Engineering in 2020 with a thesis on the performance of AlphaZero in different scenarios.
He also received the Master’s Degree in Data Science for management in 2021 with a thesis on Federated Transfer Learning.
We (me, Linara Adilova and Michael Kamp from TU Dortmund) are pleased to announce that we have a Special Issue on the Machine Learning journal on Advances in Federated Learning for Critical Applications.
Visit SI’s page for more information!
[17.03.2026] Certification announcement!
I have been certified to teach Data Parallelism: How to Train Deep Learning Models on Multiple GPUs from the NVIDIA Deep Learning Institute (DLI). I will give my first workshop on May 8th 2026. Registration is necessary.
Visit workshop’s page for more information!
[17.03.2026] Workshop announcement!
I am glad to announce that we (Christoph Düsing from the University of Bielefeld, Ioanna Miliou fmor the University fo Stockholm, Björn Eskofier from LMU Munich, and Mirko Polato, and I from UniTO) are organizing the 1st International Workshop on MULTIcentrics and Privacy-preserving Learning in Healthcare (MULTIPLi), in conjunction with the 24th International Conference on Artificial Intelligence in Medicine (AIME26), located in Ottawa, Canada.
Visit MULTIPLi’s website for more information!
[29.08.2025] Workshop announcement!
I am glad to announce that we (Linara Adilova and Michael Kamp from TU Dortmund, and Samuele Fonio, Mirko Polato, and I from UniTO) are organizing a workshop on Federated Learning for Critical Applications (FLCA), in conjunction with the 40th Annual AAAI Conference on Artificial Intelligence (AAAI26).
Visit FLCA’s website for more information!
Available Theses
Please, take a look at Theses for more info on available thesis projects.
Selected publications
[2025] Fed2RC: Federated Rocket Kernels and Ridge Classifier for Time Series Classification
Bruno Casella, Samuele Fonio, Lorenzo Sciandra, and Claudio Gallicchio, Marco Aldinucci, Mirko Polato, Roberto Esposito. In The European Conference on Artificial Intelligence, ECAI 2025, Bologna, Italy, October 25-30, 2025, 2025. DOI: 10.3233/FAIA251198
[2024] Federated Learning in a Semi-Supervised Environment for Earth Observation Data
Bruno Casella, Alessio Barbaro Chisari, Marco Aldinucci, and Sebastiano Battiato, Mario Valerio Giuffrida. In Proceedings of the 32nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN, 2024. DOI: 10.14428/esann/2024.ES2024-214
[2024] Federated Time Series Classification with ROCKET features
Bruno Casella, Jakobs Matthias, Marco Aldinucci, and Sebastian Buschjager. In Proceedings of the 32nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN, 2024. DOI: 10.14428/esann/2024.ES2024-61
[2024] FedER: Federated Learning through Experience Replay and Privacy-Preserving Data Synthesis
Matteo Pennisi, Federica Proietto Salanitri, Giovanni Bellitto, and Bruno Casella, Marco Aldinucci, Simone Palazzo, Concetto Spampinato
Computer Vision and Image Understanding, 2024. DOI: 10.1016/j.cviu.2023.103882
[2023] Federated Learning meets HPC and cloud
Iacopo Colonnelli, Bruno Casella, Gianluca Mittone, and Yasir Arfat, Barbara Cantalupo, Roberto Esposito, Alberto Riccardo Martinelli, Doriana Medić, Marco Aldinucci. In Astrophysics and Space Science Proceedings, 2023. DOI: 10.1007/978-3-031-34167-0_39
Bruno Casella, Samuele Fonio, Lorenzo Sciandra, Marco Aldinucci, Claudio Gallicchio, Mirko Polato, Roberto Esposito
Data Mining and Knowledge Discovery
ai epi eupilot fl icsc
Fed2RC is a fully federated, gradient-free method for Time Series Classification (TSC) that leverages Rocket kernels and the exact aggregation of ridge regression statistics through the closed-form solution. We propose two extended versions of Fed2RC: Fed2RC+ and MiniFed2RC. Fed2RC+ is more accurate than Fed2RC in the multiclass scenario as it enhances the feature selection step via a confidence score for each one-versus-rest classifier. MiniFed2RC employs MiniRocket, a fast and nearly deterministic version of Rocket, for feature extraction by averaging the biases, the sole source of randomness. Extensive experiments on the UCR benchmark demonstrate that Fed2RC+ and MiniFed2RC achieve state-of-the-art results, under both i.i.d. and non-i.i.d. distributions, cross-silo and cross-device scenarios, and with low computation and communication costs. Code to reproduce the experiments can be found at: https://github.com/CasellaJr/Fed2RC.
@article{casella2026minifed2rc,
title = {Fast Federated Time Series Classification using Rocket Kernels and Closed-form Classifiers},
author = {Casella, Bruno and Fonio, Samuele and Sciandra, Lorenzo and Aldinucci, Marco and Gallicchio, Claudio and Polato, Mirko and Esposito, Roberto},
year = 2026,
journal = {Data Mining and Knowledge Discovery},
%%volume = {TODO},
%%pages = {TODO--TODO},
%%doi = {TODO},
%%url = {TODO},
abstract = {Fed2RC is a fully federated, gradient-free method for Time Series Classification (TSC) that leverages Rocket kernels and the exact aggregation of ridge regression statistics through the closed-form solution. We propose two extended versions of Fed2RC: Fed2RC+ and MiniFed2RC. Fed2RC+ is more accurate than Fed2RC in the multiclass scenario as it enhances the feature selection step via a confidence score for each one-versus-rest classifier. MiniFed2RC employs MiniRocket, a fast and nearly deterministic version of Rocket, for feature extraction by averaging the biases, the sole source of randomness. Extensive experiments on the UCR benchmark demonstrate that Fed2RC+ and MiniFed2RC achieve state-of-the-art results, under both i.i.d. and non-i.i.d. distributions, cross-silo and cross-device scenarios, and with low computation and communication costs. Code to reproduce the experiments can be found at: https://github.com/CasellaJr/Fed2RC.},
keywords = {ai, fl, icsc, epi, eupilot},
%%bdsk-url-1 = {TODO},
}
The 5th Italian Conference on Big Data and Data Science - ITADATA2026
ai epi eupilot icsc
Time Series Classification (TSC) is a pivotal task in modern machine learning, with widespread applications in sensitive domains such as healthcare, finance, and cybersecurity. In real-world scenarios, data are inherently distributed and cannot be centralized due to strict privacy regulations, making federated learning a natural and necessary paradigm for collaborative model training. While several TSC methods have achieved strong performance in centralized settings, their gradient-based nature often results in prohibitive communication costs and limited efficiency when deployed in federated environments. In this context, MiniFed2RC has recently emerged as a fully federated, gradient-free approach for univariate TSC, combining random Rocket kernels with a closed-form ridge regression classifier. It has demonstrated state-of-the-art accuracy on the UCR univariate benchmark while significantly reducing computation and communication overhead. However, its effectiveness on multivariate time series—which are increasingly common in practice—has not yet been explored. In this paper, we present the first comprehensive benchmark of MiniFed2RC on the UEA multivariate time series classification archive. We evaluate the method across all 30 datasets, analyzing its predictive performance and communication costs. Our results show that MiniFed2RC achieves competitive classification performance on multivariate data, maintaining the same efficiency and robustness observed in the univariate setting. Code is available at https://github.com/CasellaJr/Fed2RC.
@inproceedings{casella2026multivarfed2rc,
title = {Experimenting with Federated Learning for Multivariate Time Series Classification},
author = {Casella, Bruno and Sciandra, Lorenzo},
year = 2026,
optmonth = {11},
booktitle = {The 5th Italian Conference on Big Data and Data Science - ITADATA2026},
pages = {1--12},
%%doi = {TODO},
%%url = {TODO},
abstract = {Time Series Classification (TSC) is a pivotal task in modern machine learning, with widespread applications in sensitive domains such as healthcare, finance, and cybersecurity. In real-world scenarios, data are inherently distributed and cannot be centralized due to strict privacy regulations, making federated learning a natural and necessary paradigm for collaborative model training. While several TSC methods have achieved strong performance in centralized settings, their gradient-based nature often results in prohibitive communication costs and limited efficiency when deployed in federated environments. In this context, MiniFed2RC has recently emerged as a fully federated, gradient-free approach for univariate TSC, combining random Rocket kernels with a closed-form ridge regression classifier. It has demonstrated state-of-the-art accuracy on the UCR univariate benchmark while significantly reducing computation and communication overhead. However, its effectiveness on multivariate time series—which are increasingly common in practice—has not yet been explored. In this paper, we present the first comprehensive benchmark of MiniFed2RC on the UEA multivariate time series classification archive. We evaluate the method across all 30 datasets, analyzing its predictive performance and communication costs. Our results show that MiniFed2RC achieves competitive classification performance on multivariate data, maintaining the same efficiency and robustness observed in the univariate setting. Code is available at https://github.com/CasellaJr/Fed2RC.},
keywords = {ai, epi, icsc, eupilot},
}
Luca Grandi, Loris Signoretti, Lorenzo Sciandra, Bruno Casella, Marco Aldinucci, Roberto Esposito
35th International Conference on Artificial Neural Networks – ICANN 2026
ai epi eupilot icsc
@inproceedings{Grandi2026Roc2Img,
title = {A Lightweight Random Feature Framework for Efficient Image Classification},
author = {Grandi, Luca and Signoretti, Loris and Sciandra, Lorenzo and Casella, Bruno and Aldinucci, Marco and Esposito, Roberto},
year = 2026,
month = {09},
booktitle = {35th International Conference on Artificial Neural Networks – ICANN 2026},
address = {Padova, Italy},
pages = {1--12},
doi = {10.1007/978-3-032-38401-0_54},
url = {https://iris.unito.it/retrieve/e5f10488-0568-4d6c-ada7-692b710240db/_ICANN_26__Roc2Imgs.pdf},
keywords = {ai, epi, icsc, eupilot},
}
Proceedings of the ECML-PKDD Workshop, 3nd workshop on advancements in Federated Learning
epi icsc
Federated Learning (FL) systems face numerous challenges, including data heterogeneity, communication constraints, and vulnerability to adversarial attacks. A promising approach that has recently gained attention is prototype learning, which gives the possibility to share only local prototypes rather than entire models. However, the reliability of such systems is often overlooked, leading to insufficient attention to potential attack vectors. In this paper, we investigate the robustness of two Federated Prototypical methods under both random and revert poisoning attacks. We propose an extension of these methods using personalized aggregation in two different ways to enhance their resilience. Our approach provides valuable insights into the reliability of Federated Prototype Learning and contributes to the development of more robust FL systems. We validate our findings through an extensive experimental evaluation on Cifar-10 and EMNIST, with gains in performance up to 10
@inproceedings{fonio2025perfed,
abstract = {Federated Learning (FL) systems face numerous challenges, including data heterogeneity, communication constraints, and vulnerability to adversarial attacks. A promising approach that has recently gained attention is prototype learning, which gives the possibility to share only local prototypes rather than entire models. However, the reliability of such systems is often overlooked, leading to insufficient attention to potential attack vectors. In this paper, we investigate the robustness of two Federated Prototypical methods under both random and revert poisoning attacks. We propose an extension of these methods using personalized aggregation in two different ways to enhance their resilience. Our approach provides valuable insights into the reliability of Federated Prototype Learning and contributes to the development of more robust FL systems. We validate our findings through an extensive experimental evaluation on Cifar-10 and EMNIST, with gains in performance up to 10% and 20%, respectively, with respect to the baselines.},
author = {Fonio, Samuele and Casella, Bruno and Aldinucci, Marco},
booktitle = {Proceedings of the ECML-PKDD Workshop, 3nd workshop on advancements in Federated Learning},
keywords = {epi, icsc},
location = {Porto, Portugal},
title = {{Personalized Aggregation for Federated Prototypical Learning}},
url = {https://iris.unito.it/retrieve/47a8316b-62b7-4142-a0cc-a867d37175e3/wafl_preprint.pdf},
year = {2025},
bdsk-url-1 = {https://iris.unito.it/retrieve/47a8316b-62b7-4142-a0cc-a867d37175e3/wafl_preprint.pdf}}
Bruno Casella, Matthias Jakobs, Marco Aldinucci, Sebastian Buschjaeger
Proceedings of the ECML-PKDD Workshop, 3nd workshop on advancements in Federated Learning
epi icsc
Time series classification (TSC) is a critical task with applications in various domains, including healthcare, finance, and industrial monitoring. Due to privacy concerns and data regulations, Federated Learning has emerged as a promising approach for learning from distributed time series data without centralizing raw information. However, most FL solutions rely on a client-server architecture, which introduces robustness and confidentiality risks related to the distinguished role of the server, which is a single point of failure and can observe knowledge extracted from clients. To address these challenges, we propose DROCKS, a fully decentralized FL framework for TSC that leverages ROCKET (RandOm Convolutional KErnel Transform) features. In DROCKS, the global model is trained by sequentially traversing a structured path across federation nodes, where each node refines the model and selects the most effective local kernels before passing them to the successor. Extensive experiments on the UCR archive demonstrate that DROCKS outperforms state-of-the-art client-server FL approaches while being more resilient to node failures and malicious attacks. Our code is available at https://anonymous.4open.science/r/DROCKS-7FF3/README.md.
@inproceedings{casella2025drocks,
abstract = {Time series classification (TSC) is a critical task with applications in various domains, including healthcare, finance, and industrial monitoring. Due to privacy concerns and data regulations, Federated Learning has emerged as a promising approach for learning from distributed time series data without centralizing raw information. However, most FL solutions rely on a client-server architecture, which introduces robustness and confidentiality risks related to the distinguished role of the server, which is a single point of failure and can observe knowledge extracted from clients. To address these challenges, we propose DROCKS, a fully decentralized FL framework for TSC that leverages ROCKET (RandOm Convolutional KErnel Transform) features. In DROCKS, the global model is trained by sequentially traversing a structured path across federation nodes, where each node refines the model and selects the most effective local kernels before passing them to the successor. Extensive experiments on the UCR archive demonstrate that DROCKS outperforms state-of-the-art client-server FL approaches while being more resilient to node failures and malicious attacks. Our code is available at https://anonymous.4open.science/r/DROCKS-7FF3/README.md.},
author = {Casella, Bruno and Jakobs, Matthias and Aldinucci, Marco and Buschjaeger, Sebastian},
booktitle = {Proceedings of the ECML-PKDD Workshop, 3nd workshop on advancements in Federated Learning},
keywords = {epi, icsc},
location = {Porto, Portugal},
title = {{Decentralized Time Series Classification with ROCKET Features}},
url = {https://iris.unito.it/retrieve/b3d4f7f0-c7ec-4ad3-96be-6b55add95b6a/FAIA-413-FAIA251198.pdf},
year = {2025},
bdsk-url-1 = {https://iris.unito.it/retrieve/b3d4f7f0-c7ec-4ad3-96be-6b55add95b6a/FAIA-413-FAIA251198.pdf}}
Bruno Casella, Samuele Fonio, Lorenzo Sciandra, Marco Aldinucci, Claudio Gallicchio, Mirko Polato, Roberto Esposito
28th European Conference on Artificial Intelligence, 25-30 October 2025, Bologna, Italy – Including 14th Conference on Prestigious Applications of Intelligent Systems (PAIS 2025)
epi fl icsc
Time series classification is a pivotal task in modern machine learning, with widespread applications in fields such as healthcare, finance, and cybersecurity. While deep learning methods dominate recent developments, their resource demands and privacy limitations hinder deployment on low-power and decentralized environments. To address these challenges, we introduce Fed2RC, a fully federated and gradient-free approach that integrates the efficiency of Rocket-based feature extraction with the robustness of ridge regression in a privacy-preserving setting. Fed2RC builds upon two key ideas: (i) federated selection and aggregation of high-performing random convolution kernels, and (ii) incremental and communication-efficient updates of ridge classifier parameters using closed-form solutions. Additionally, we propose a novel federated protocol for selecting the global ridge regularization parameter λ, and show how to improve the communication efficiency by matrix factorization techniques. Extensive experiments on the UCR benchmark demonstrate that Fed2RC achieves state-of-the-art results with a fraction of the computation and communication costs. Code to reproduce the experiments can be found at: https://github.com/CasellaJr/Fed2RC.
@inproceedings{25:casella:fed2rc,
abstract = {Time series classification is a pivotal task in modern machine learning, with widespread applications in fields such as healthcare, finance, and cybersecurity. While deep learning methods dominate recent developments, their resource demands and privacy limitations hinder deployment on low-power and decentralized environments. To address these challenges, we introduce Fed2RC, a fully federated and gradient-free approach that integrates the efficiency of Rocket-based feature extraction with the robustness of ridge regression in a privacy-preserving setting. Fed2RC builds upon two key ideas: (i) federated selection and aggregation of high-performing random convolution kernels, and (ii) incremental and communication-efficient updates of ridge classifier parameters using closed-form solutions. Additionally, we propose a novel federated protocol for selecting the global ridge regularization parameter λ, and show how to improve the communication efficiency by matrix factorization techniques. Extensive experiments on the UCR benchmark demonstrate that Fed2RC achieves state-of-the-art results with a fraction of the computation and communication costs. Code to reproduce the experiments can be found at: https://github.com/CasellaJr/Fed2RC.},
author = {Casella, Bruno and Fonio, Samuele and Sciandra, Lorenzo and Aldinucci, Marco and Gallicchio, Claudio and Polato, Mirko and Esposito, Roberto},
booktitle = {28th European Conference on Artificial Intelligence, 25-30 October 2025, Bologna, Italy -- Including 14th Conference on Prestigious Applications of Intelligent Systems (PAIS 2025)},
doi = {https://dx.doi.org/10.3233/faia251198},
keywords = {icsc, epi, fl},
month = oct,
pages = {3297--3304},
publisher = {IOS Press},
title = {Fed2RC: Federated Rocket Kernels and Ridge Classifier for Time Series Classification},
url = {https://ebooks.iospress.nl/doi/10.3233/FAIA251198},
year = {2025},
bdsk-url-1 = {https://ebooks.iospress.nl/doi/10.3233/FAIA251198},
bdsk-url-2 = {https://dx.doi.org/10.3233/faia251198}}
Alessio Barbaro Chisari, Luca Guarnera, Alessandro Ortis, Wladimiro Carlo Patatu, Bruno Casella, Luca Naso, Giuseppe Puglisi, Vincenzo Del Zoppo, Mario Valerio Giuffrida, Sebastiano Battiato
IEEE Access
ai
Accurate cloud detection is critical for advancing atmospheric monitoring and meteorological forecasting. This paper presents the Cloud Detection Challenge, an initiative aimed at enhancing cloud detection through innovative solutions using lidar-based ceilometer data. This initiative was hosted by IEEE MetroXRAINE 2024, and 11 teams participated in this initiative. Participants were provided with a novel dataset of backscatter profiles converted into time-height plots, offering unique insights into atmospheric conditions beyond conventional imagery. Data collection employed a Lufft CHM 15k ceilometer, capturing cloud dynamics every 15 seconds located near Mt. Etna, an active volcano in Italy. The dataset includes 1568 hourly labeled backscatter profiles, serving as a benchmark for state-of-the-art deep learning models. The challenge sets a baseline performance of 89.57% accuracy, 92.73% F1-score, 89.82% precision, and 95.84% recall, inviting participants to develop models to exceed these results. Submissions proposed a widerange of AI-based approaches, including Transformer and Convolutional Neural Network architectures, showcasing the potential of advanced image analysis techniques in lidar-based cloud detection. This paper details the challenge framework, as well as the methodologies proposed by top-performing teams, offering a comparative evaluation of their effectiveness. Our initiative advances cloud detection technologies and underscores their broader implications for environmental monitoring, agriculture, and satellite imaging. The insights and dataset presented herein lay the groundwork for future advancements in leveraging lidar data for atmospheric analysis.
@article{chisari2025cloud,
abstract = {Accurate cloud detection is critical for advancing atmospheric monitoring and meteorological forecasting. This paper presents the Cloud Detection Challenge, an initiative aimed at enhancing cloud detection through innovative solutions using lidar-based ceilometer data. This initiative was hosted by IEEE MetroXRAINE 2024, and 11 teams participated in this initiative. Participants were provided with a novel dataset of backscatter profiles converted into time-height plots, offering unique insights into atmospheric conditions beyond conventional imagery. Data collection employed a Lufft CHM 15k ceilometer, capturing cloud dynamics every 15 seconds located near Mt. Etna, an active volcano in Italy. The dataset includes 1568 hourly labeled backscatter profiles, serving as a benchmark for state-of-the-art deep learning models. The challenge sets a baseline performance of 89.57\% accuracy, 92.73\% F1-score, 89.82\% precision, and 95.84\% recall, inviting participants to develop models to exceed these results. Submissions proposed a widerange of AI-based approaches, including Transformer and Convolutional Neural Network architectures, showcasing the potential of advanced image analysis techniques in lidar-based cloud detection. This paper details the challenge framework, as well as the methodologies proposed by top-performing teams, offering a comparative evaluation of their effectiveness. Our initiative advances cloud detection technologies and underscores their broader implications for environmental monitoring, agriculture, and satellite imaging. The insights and dataset presented herein lay the groundwork for future advancements in leveraging lidar data for atmospheric analysis.},
author = {Barbaro Chisari, Alessio and Guarnera, Luca and Ortis, Alessandro and Patatu, Wladimiro Carlo and Casella, Bruno and Naso, Luca and Puglisi, Giuseppe and Del Zoppo, Vincenzo and Giuffrida, Mario Valerio and Battiato, Sebastiano},
date-modified = {2025-05-24 09:17:18 +0200},
doi = {10.1109/ACCESS.2025.3553422},
journal = {IEEE Access},
keywords = {ai},
pages = {53394--53406},
title = {Cloud Detection Challenge - Methods and Results},
url = {https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10934992},
volume = {13},
year = {2025},
bdsk-url-1 = {https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10934992}}
2025 33rd Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP)
epi icsc
Federated Learning (FL) is a distributed machine learning approach that has emerged as an effective way to address recent privacy concerns. However, FL introduces the need for additional security measures as FL alone is still subject to vulner- abilities such as model and data poisoning and inference attacks. Confidential Computing (CC) is a paradigm that, by leveraging hardware-based trusted execution environments (TEEs), protects the confidentiality and integrity of ML models and data, thus resulting in a powerful ally of FL applications. Typical TEEs offer an application-isolation level but suffer many drawbacks, such as limited available memory and debugging and coding difficulties. The new generation of TEEs offers a virtual machine (VM)- based isolation level, thus reducing the porting effort for existing applications. In this work, we compare the performance of VM- based and application-isolation level TEEs for confidential FL (CFL) applications. In particular, we evaluate the impact of TEEs and additional security mechanisms such as TLS (for securing the communication channel). The results, obtained across three datasets and two deep learning models, demonstrate that the new VM-based TEEs introduce a limited overhead (at most 1.5x), thus paving the way to leverage public and untrusted computing environments, such as HPC facilities or public cloud, without detriment to performance.
@article{casella2025performance,
abstract = {Federated Learning (FL) is a distributed machine learning approach that has emerged as an effective way to address recent privacy concerns. However, FL introduces the need for additional security measures as FL alone is still subject to vulner- abilities such as model and data poisoning and inference attacks. Confidential Computing (CC) is a paradigm that, by leveraging hardware-based trusted execution environments (TEEs), protects the confidentiality and integrity of ML models and data, thus resulting in a powerful ally of FL applications. Typical TEEs offer an application-isolation level but suffer many drawbacks, such as limited available memory and debugging and coding difficulties. The new generation of TEEs offers a virtual machine (VM)- based isolation level, thus reducing the porting effort for existing applications. In this work, we compare the performance of VM- based and application-isolation level TEEs for confidential FL (CFL) applications. In particular, we evaluate the impact of TEEs and additional security mechanisms such as TLS (for securing the communication channel). The results, obtained across three datasets and two deep learning models, demonstrate that the new VM-based TEEs introduce a limited overhead (at most 1.5x), thus paving the way to leverage public and untrusted computing environments, such as HPC facilities or public cloud, without detriment to performance.},
address = {Turin, Italy},
author = {Bruno Casella},
booktitle = {2025 33rd Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP)},
doi = {10.1109/PDP66500.2025.00035},
keywords = {icsc, epi},
pages = {204--208},
title = {A performance analysis of VM-based Trusted Execution Environments for Confidential Federated Learning},
url = {https://iris.unito.it/retrieve/4cd172c3-dee9-4c0c-9f34-7bcbae584571/PDP25___A_performance_comparison_of_Confidential_Computing_architectures_for_Federated_Learning_tasks.pdf},
year = {2025},
bdsk-url-1 = {https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10974758},
bdsk-url-2 = {https://doi.org/10.1109/PDP66500.2025.00035}}
Oussama Harrak, Bruno Casella, Samuele Fonio, Piero Fariselli, Gianluca Mittone, Tiziana Sanavia, Marco Aldinucci
Proceedings of the ECML-PKDD Workshop, 2nd workshop on advancements in Federated Learning
ai epi fl icsc
This work proposes FedSurvBoost, a federated learning pipeline for survival analysis based on the AdaBoost.F algorithm, which iteratively aggregates the best local weak hypotheses. Our method extends AdaBoost.F by removing the dependence on the number of classes coefficient from the computation of the weights of the best model. This makes it suitable for regression tasks, such as survival analysis. We show the effectiveness of our approach by comparing it with state-of-the-art methods, specifically developed for survival analysis problems, on two common survival datasets. Our code is available at https://github.com/oussamaHarrak/FedSurvBoost.
@inproceedings{harrak2024fedsurvboost,
abstract = {This work proposes FedSurvBoost, a federated learning pipeline for survival analysis based on the AdaBoost.F algorithm, which iteratively aggregates the best local weak hypotheses. Our method extends AdaBoost.F by removing the dependence on the number of classes coefficient from the computation of the weights of the best model. This makes it suitable for regression tasks, such as survival analysis. We show the effectiveness of our approach by comparing it with state-of-the-art methods, specifically developed for survival analysis problems, on two common survival datasets. Our code is available at https://github.com/oussamaHarrak/FedSurvBoost.},
author = {Harrak, Oussama and Casella, Bruno and Fonio, Samuele and Fariselli, Piero and Mittone, Gianluca and Sanavia, Tiziana and Aldinucci, Marco},
booktitle = {Proceedings of the ECML-PKDD Workshop, 2nd workshop on advancements in Federated Learning},
keywords = {epi, icsc, fl, ai},
location = {Vilnius, Lithuania},
title = {{Federated AdaBoost for Survival Analysis}},
url = {https://iris.unito.it/retrieve/9d7ca2cc-ce81-4784-90cf-c5b5cf9739fb/WAFL___ECMLPKDD___FedSurvBoost.pdf},
year = {2024},
bdsk-url-1 = {https://iris.unito.it/retrieve/9d7ca2cc-ce81-4784-90cf-c5b5cf9739fb/WAFL___ECMLPKDD___FedSurvBoost.pdf}}
Bruno Casella, Alessio Barbaro Chisari, Marco Aldinucci, Sebastiano Battiato, Mario Valerio Giuffrida
Proceedings of the 32nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN
ai epi icsc
We propose FedRec, a federated learning workflow taking advantage of unlabelled data in a semi-supervised environment to assist in the training of a supervised aggregated model. In our proposed method, an encoder architecture extracting features from unlabelled data is aggregated with the feature extractor of a classification model via weight averaging. The fully connected layers of the supervised models are also averaged in a federated fashion. We show the effectiveness of our approach by comparing it with the state-of-the-art federated algorithm, an isolated and a centralised baseline, on novel cloud detection datasets.
@inproceedings{24:casella:fedrec,
abstract = {We propose FedRec, a federated learning workflow taking advantage of unlabelled data in a semi-supervised environment to assist in the training of a supervised aggregated model. In our proposed method, an encoder architecture extracting features from unlabelled data is aggregated with the feature extractor of a classification model via weight averaging. The fully connected layers of the supervised models are also averaged in a federated fashion. We show the effectiveness of our approach by comparing it with the state-of-the-art federated algorithm, an isolated and a centralised baseline, on novel cloud detection datasets.},
author = {Casella, Bruno and Chisari, Alessio Barbaro and Aldinucci, Marco and Battiato, Sebastiano and Giuffrida, Mario Valerio},
booktitle = {Proceedings of the 32nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, {ESANN}},
doi = {10.14428/esann/2024.es2024-214},
keywords = {icsc, epi, ai},
location = {Bruges, Belgium},
month = oct,
title = {Federated Learning in a Semi-Supervised Environment for Earth Observation Data},
url = {https://iris.unito.it/retrieve/a798d7b8-6b98-48c2-92f4-327d2aaa8788/ES2024-214.pdf},
year = {2024},
bdsk-url-1 = {https://iris.unito.it/retrieve/a798d7b8-6b98-48c2-92f4-327d2aaa8788/ES2024-214.pdf}}
Bruno Casella, Jakobs Matthias, Marco Aldinucci, Sebastian Buschjager
Proceedings of the 32nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN
ai epi icsc
This paper proposes FROCKS, a federated time series classification method using ROCKET features. Our approach dynamically adapts the models' features by selecting and exchanging the best-performing ROCKET kernels from a federation of clients. Specifically, the server gathers the best-performing kernels of the clients together with the associated model parameters, and it performs a weighted average if a kernel is best-performing for more than one client. We compare the proposed method with state-of-the-art approaches on the UCR archive binary classification datasets and show superior performance on most datasets.
@inproceedings{24:casella:frocks,
abstract = {This paper proposes FROCKS, a federated time series classification method using ROCKET features. Our approach dynamically adapts the models' features by selecting and exchanging the best-performing ROCKET kernels from a federation of clients. Specifically, the server gathers the best-performing kernels of the clients together with the associated model parameters, and it performs a weighted average if a kernel is best-performing for more than one client. We compare the proposed method with state-of-the-art approaches on the UCR archive binary classification datasets and show superior performance on most datasets.},
author = {Casella, Bruno and Matthias, Jakobs and Aldinucci, Marco and Buschjager, Sebastian},
booktitle = {Proceedings of the 32nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, {ESANN}},
doi = {10.14428/esann/2024.es2024-61},
keywords = {icsc, epi, ai},
location = {Bruges, Belgium},
month = oct,
title = {Federated Time Series Classification with ROCKET features},
url = {https://iris.unito.it/retrieve/51b63fc1-3e22-4ad4-8926-84af69cde739/ES2024-61.pdf},
year = {2024},
bdsk-url-1 = {https://iris.unito.it/retrieve/51b63fc1-3e22-4ad4-8926-84af69cde739/ES2024-61.pdf}}
Bruno Casella, Roberto Esposito, Antonio Sciarappa, Carlo Cavazzoni, Marco Aldinucci
IEEE Access
ai epi icsc
@article{24:casella:normalization,
author = {Casella, Bruno and Esposito, Roberto and Sciarappa, Antonio and Cavazzoni, Carlo and Aldinucci, Marco},
date-modified = {2025-09-23 17:48:26 +0200},
doi = {10.1109/ACCESS.2024.3383783},
journal = {IEEE Access},
keywords = {epi, icsc, ai},
pages = {47961--47971},
title = {Experimenting With Normalization Layers in Federated Learning on Non-IID Scenarios},
volume = {12},
year = {2024},
bdsk-url-1 = {https://doi.org/10.1109/ACCESS.2024.3383783}}
Bruno Casella, Iacopo Colonnelli, Gianluca Mittone, Robert Birke, Walter Riviera, Antonio Sciarappa, Carlo Cavazzoni, Marco Aldinucci
Proceedings of the 2024 Deep Learning Security and Privacy Workshop, IEEE Symposium on Security and Privacy 2024
confidential epi fl icsc
Federated Learning (FL) has emerged as a solution to preserve data privacy by keeping the data locally on each participant's device. However, FL alone is still vulnerable to attacks that can cause privacy leaks. Therefore, it becomes necessary to take additional security measures at the cost of increasing runtimes. The Trusted Execution Environment (TEE) approach promises to offer the highest degree of security during execution. However, TEEs suffer from memory limits which prevent safe end-to-end FL training of modern deep models. State-of- the-art approaches limit secure training to selected layers, failing to avert the full spectrum of attacks or adopt layer-wise training affecting model performance. We benchmark the usage of a library OS (LibOS) to run the full, unmodified end-to-end FL training inside the TEE. We extensively evaluate and model the overhead of the different security mechanisms needed to protect the data and model during computation (TEE), communication (TLS), and storage (disk encryption). The obtained results across three datasets and two models demonstrate that LibOSes are a viable way to seamlessly inject security into FL with limited overhead (at most 2x), offering valuable guidance for researchers and developers aiming to apply FL in data-security-focused contexts.
@inproceedings{24:casella:sgx,
abstract = {Federated Learning (FL) has emerged as a solution to preserve data privacy by keeping the data locally on each participant's device. However, FL alone is still vulnerable to attacks that can cause privacy leaks. Therefore, it becomes necessary to take additional security measures at the cost of increasing runtimes. The Trusted Execution Environment (TEE) approach promises to offer the highest degree of security during execution. However, TEEs suffer from memory limits which prevent safe end-to-end FL training of modern deep models. State-of- the-art approaches limit secure training to selected layers, failing to avert the full spectrum of attacks or adopt layer-wise training affecting model performance. We benchmark the usage of a library OS (LibOS) to run the full, unmodified end-to-end FL training inside the TEE. We extensively evaluate and model the overhead of the different security mechanisms needed to protect the data and model during computation (TEE), communication (TLS), and storage (disk encryption). The obtained results across three datasets and two models demonstrate that LibOSes are a viable way to seamlessly inject security into FL with limited overhead (at most 2x), offering valuable guidance for researchers and developers aiming to apply FL in data-security-focused contexts.},
author = {Casella, Bruno and Colonnelli, Iacopo and Mittone, Gianluca and Birke, Robert and Riviera, Walter and Sciarappa, Antonio and Cavazzoni, Carlo and Aldinucci, Marco},
booktitle = {Proceedings of the 2024 Deep Learning Security and Privacy Workshop, IEEE Symposium on Security and Privacy 2024},
date-modified = {2024-08-18 13:16:08 +0200},
doi = {10.1109/SPW63631.2024.00009},
keywords = {icsc, epi, confidential, fl},
location = {San Francisco, CA},
month = may,
title = {A Performance Analysis for Confidential Federated Learning},
url = {https://iris.unito.it/retrieve/b5877a97-2d8d-4e95-8791-0aa4a1b953b3/DLSP___CONFIDENTIAL_FL.pdf},
year = {2024},
bdsk-url-1 = {https://iris.unito.it/retrieve/b5877a97-2d8d-4e95-8791-0aa4a1b953b3/DLSP___CONFIDENTIAL_FL.pdf}}
Bruno Casella, Walter Riviera, Marco Aldinucci, Gloria Menegaz
STAR Protocols
epi icsc
Federated learning is a cooperative learning approach that has emerged as an effective way to address privacy concerns. Here, we present a protocol for training MERGE: a federated multi-input neural network (NN) for COVID-19 prognosis. We describe steps for collecting and preprocessing datasets. We then detail the process of training a multi-input NN. This protocol can be adapted for use with datasets containing both image- and table-based input sources.
@article{24:casella:starprotocol,
abstract = {Federated learning is a cooperative learning approach that has emerged as an effective way to address privacy concerns. Here, we present a protocol for training MERGE: a federated multi-input neural network (NN) for COVID-19 prognosis. We describe steps for collecting and preprocessing datasets. We then detail the process of training a multi-input NN. This protocol can be adapted for use with datasets containing both image- and table-based input sources.},
author = {Casella, Bruno and Riviera, Walter and Aldinucci, Marco and Menegaz, Gloria},
date-modified = {2024-04-01 11:14:16 +0200},
doi = {10.1016/j.xpro.2023.102812},
institution = {Computer Science Department, University of Torino},
journal = {STAR Protocols},
keywords = {icsc, epi},
note = {https://prod-shared-star-protocols.s3.amazonaws.com/protocols/3225.pdf},
title = {Protocol for training MERGE: A federated multi-input neural network for COVID-19 prognosis},
url = {https://prod-shared-star-protocols.s3.amazonaws.com/protocols/3225.pdf},
year = {2024},
bdsk-url-1 = {https://prod-shared-star-protocols.s3.amazonaws.com/protocols/3225.pdf},
bdsk-url-2 = {https://doi.org/10.1016/j.xpro.2023.102812}}
Matteo Pennisi, Federica Proietto Salanitri, Giovanni Bellitto, Bruno Casella, Marco Aldinucci, Simone Palazzo, Concetto Spampinato
Computer Vision and Image Understanding
ai
In the medical field, multi-center collaborations are often sought to yield more generalizable findings by leveraging the heterogeneity of patient and clinical data. However, recent privacy regulations hinder the possibility to share data, and consequently, to come up with machine learning-based solutions that support diagnosis and prognosis. Federated learning (FL) aims at sidestepping this limitation by bringing AI-based solutions to data owners and only sharing local AI models, or parts thereof, that need then to be aggregated. However, most of the existing federated learning solutions are still at their infancy and show several shortcomings, from the lack of a reliable and effective aggregation scheme able to retain the knowledge learned locally to weak privacy preservation as real data may be reconstructed from model updates. Furthermore, the majority of these approaches, especially those dealing with medical data, relies on a centralized distributed learning strategy that poses robustness, scalability and trust issues. In this paper we present a federated and decentralized learning strategy, FedER, that, exploiting experience replay and generative adversarial concepts, effectively integrates features from local nodes, providing models able to generalize across multiple datasets while maintaining privacy. FedER is tested on two tasks — tuberculosis and melanoma classification — using multiple datasets in order to simulate realistic non-i.i.d. medical data scenarios. Results show that our approach achieves performance comparable to standard (non-federated) learning and significantly outperforms state-of-the-art federated methods in their centralized (thus, more favourable) formulation. Code is available at https://github.com/perceivelab/FedER
@article{23:casella:FedER,
abstract = {In the medical field, multi-center collaborations are often sought to yield more generalizable findings by leveraging the heterogeneity of patient and clinical data. However, recent privacy regulations hinder the possibility to share data, and consequently, to come up with machine learning-based solutions that support diagnosis and prognosis. Federated learning (FL) aims at sidestepping this limitation by bringing AI-based solutions to data owners and only sharing local AI models, or parts thereof, that need then to be aggregated. However, most of the existing federated learning solutions are still at their infancy and show several shortcomings, from the lack of a reliable and effective aggregation scheme able to retain the knowledge learned locally to weak privacy preservation as real data may be reconstructed from model updates. Furthermore, the majority of these approaches, especially those dealing with medical data, relies on a centralized distributed learning strategy that poses robustness, scalability and trust issues. In this paper we present a federated and decentralized learning strategy, FedER, that, exploiting experience replay and generative adversarial concepts, effectively integrates features from local nodes, providing models able to generalize across multiple datasets while maintaining privacy. FedER is tested on two tasks --- tuberculosis and melanoma classification --- using multiple datasets in order to simulate realistic non-i.i.d. medical data scenarios. Results show that our approach achieves performance comparable to standard (non-federated) learning and significantly outperforms state-of-the-art federated methods in their centralized (thus, more favourable) formulation. Code is available at https://github.com/perceivelab/FedER},
author = {Pennisi, Matteo and Proietto Salanitri, Federica and Bellitto, Giovanni and Casella, Bruno and Aldinucci, Marco and Palazzo, Simone and Spampinato, Concetto},
date-modified = {2024-04-01 11:13:27 +0200},
doi = {10.1016/j.cviu.2023.103882},
institution = {Computer Science Department, University of Torino},
journal = {Computer Vision and Image Understanding},
keywords = {ai},
optnote = {https://www.sciencedirect.com/science/article/pii/S107731422300262X?via%3Dihub},
pages = {103882},
title = {FedER: Federated Learning through Experience Replay and Privacy-Preserving Data Synthesis},
url = {https://www.sciencedirect.com/science/article/pii/S107731422300262X?via%3Dihub},
volume = {238},
year = {2024},
bdsk-url-1 = {https://www.sciencedirect.com/science/article/pii/S107731422300262X?via%3Dihub},
bdsk-url-2 = {https://doi.org/10.1016/j.cviu.2023.103882}}
Proceedings of the 2023 IEEE International Conference on Blockchain and Cryptocurrency (ICBC), 1-5 May 2023, Dubai
epi icsc
Blockchain, the underlying technology of Bitcoin and several other cryptocurrencies, like Ethereum, produces a massive amount of open-access data that can be analyzed, providing important information about the network's activity and its respective token. The on-chain data have extensively been used as input to Machine Learning algorithms for predicting cryptocurrencies' future prices; however, there is a lack of study in predicting the future behaviour of on-chain data. This study aims to show how on-chain data can be used to detect cryptocurrency market regimes, like minimum and maximum, bear and bull market phases, and how forecasting these data can provide an optimal asset allocation for long-term investors.
@inproceedings{23:casella:onchain,
abstract = {Blockchain, the underlying technology of Bitcoin and several other cryptocurrencies, like Ethereum, produces a massive amount of open-access data that can be analyzed, providing important information about the network's activity and its respective token. The on-chain data have extensively been used as input to Machine Learning algorithms for predicting cryptocurrencies' future prices; however, there is a lack of study in predicting the future behaviour of on-chain data. This study aims to show how on-chain data can be used to detect cryptocurrency market regimes, like minimum and maximum, bear and bull market phases, and how forecasting these data can provide an optimal asset allocation for long-term investors.},
author = {Casella, Bruno and Paletto, Lorenzo},
booktitle = {Proceedings of the 2023 IEEE International Conference on Blockchain and Cryptocurrency (ICBC), 1-5 May 2023, Dubai},
doi = {https://doi.org/10.1109/ICBC56567.2023.10174989},
keywords = {icsc, epi},
title = {Predicting Cryptocurrencies Market Phases through On-Chain Data Long-Term Forecasting},
url = {https://iris.unito.it/bitstream/2318/1902652/1/6.%20ICBC23%20-%20PREDICTING%20BTC.pdf},
year = {2023},
bdsk-url-1 = {https://iris.unito.it/bitstream/2318/1902652/1/6.%20ICBC23%20-%20PREDICTING%20BTC.pdf},
bdsk-url-2 = {https://doi.org/10.1109/ICBC56567.2023.10174989}}
Proceedings of the 3rd Workshop on Distributed Machine Learning for the Intelligent Computing Continuum (DML-ICC), IEEE/ACM UCC 2023, Taormina, Italy, 4 December 2023
ai epi icsc
Federated Learning (FL) has emerged as a promising solution to address privacy concerns by collaboratively training Deep Learning (DL) models across distributed parties. This work proposes an architecture-based aggregation strategy in Vertical FL, where parties hold data with different attributes but shared instances. Our approach leverages the identical architectural parts, i.e. neural network layers, of different models to selectively aggregate weights, which is particularly relevant when collaborating with institutions holding different types of datasets, i.e., image, text, or tabular datasets. In a scenario where two entities train DL models, such as a Convolutional Neural Network (CNN) and a Multi-Layer Perceptron (MLP), our strategy computes the average only for architecturally identical segments. This preserves data-specific features learned from demographic and clinical data. We tested our approach on two clinical datasets, i.e., the COVID-CXR dataset and the ADNI study. Results show that our method achieves comparable results with the centralized scenario, in which all the data are collected in a single data lake, and benefits from FL generalizability. In particular, compared to the non-federated models, our proposed proof-of-concept model exhibits a slight performance loss on the COVID-CXR dataset (less than 8
@inproceedings{23:casella:architecturalfedavg,
abstract = {Federated Learning (FL) has emerged as a promising solution to address privacy concerns by collaboratively training Deep Learning (DL) models across distributed parties. This work proposes an architecture-based aggregation strategy in Vertical FL, where parties hold data with different attributes but shared instances. Our approach leverages the identical architectural parts, i.e. neural network layers, of different models to selectively aggregate weights, which is particularly relevant when collaborating with institutions holding different types of datasets, i.e., image, text, or tabular datasets. In a scenario where two entities train DL models, such as a Convolutional Neural Network (CNN) and a Multi-Layer Perceptron (MLP), our strategy computes the average only for architecturally identical segments. This preserves data-specific features learned from demographic and clinical data. We tested our approach on two clinical datasets, i.e., the COVID-CXR dataset and the ADNI study. Results show that our method achieves comparable results with the centralized scenario, in which all the data are collected in a single data lake, and benefits from FL generalizability. In particular, compared to the non-federated models, our proposed proof-of-concept model exhibits a slight performance loss on the COVID-CXR dataset (less than 8%), but outperforms ADNI models by up to 12%. Moreover, communication costs between training rounds are minimized by exchanging only the dense layer parameters.},
author = {Casella, Bruno and Fonio, Samuele},
booktitle = {Proceedings of the 3rd Workshop on Distributed Machine Learning for the Intelligent Computing Continuum (DML-ICC), IEEE/ACM UCC 2023, Taormina, Italy, 4 December 2023},
doi = {10.1145/3603166.3632559},
keywords = {icsc, epi, ai},
title = {Architecture-Based FedAvg for Vertical Federated Learning},
url = {https://iris.unito.it/retrieve/173d9960-8531-419d-9bd5-5acce6694c4e/Aggregation%20Based%20VFL.pdf},
year = {2023},
bdsk-url-1 = {https://iris.unito.it/retrieve/173d9960-8531-419d-9bd5-5acce6694c4e/Aggregation%20Based%20VFL.pdf},
bdsk-url-2 = {10.1145/3603166.3632559}}
Matteo Pennisi, Federica Proietto Salanitri, Giovanni Bellitto, Bruno Casella, Marco Aldinucci, Simone Palazzo, Concetto Spampinato
Proceedings of the 1st Workshop on Visual Continual Learning, ICCV 2023, Paris, France, 2 October 2023
ai
Federated and continual learning are training paradigms addressing data distribution shift in space and time. More specifically, federated learning tackles non-i.i.d data in space as information is distributed in multiple nodes, while continual learning faces with temporal aspect of training as it deals with continuous streams of data. Distribution shifts over space and time is what it happens in real federated learning scenarios that show multiple challenges. First, the federated model needs to learn sequentially while retaining knowledge from the past training rounds. Second, the model has also to deal with concept drift from the distributed data distributions. To address these complexities, we attempt to combine continual and federated learning strategies by proposing a solution inspired by experience replay and generative adversarial concepts for supporting decentralized distributed training. In particular, our approach relies on using limited memory buffers of synthetic privacy-preserving samples and interleaving training on local data and on buffer data. By translating the CL formulation into the task of integrating distributed knowledge with local knowledge, our method enables models to effectively integrate learned representation from local nodes, providing models the capability to generalize across multiple datasets.We test our integrated strategy on two realistic medical image analysis tasks — tuberculosis and melanoma classification — using multiple datasets in order to simulate realistic non-i.i.d. medical data scenarios. Results show that our approach achieves performance comparable to standard (non-federated) learning and significantly outperforms state-of-the-art federated methods in their centralized (thus, more favourable) formulation.
@inproceedings{23:casella:ERGANs,
abstract = {Federated and continual learning are training paradigms addressing data distribution shift in space and time. More specifically, federated learning tackles non-i.i.d data in space as information is distributed in multiple nodes, while continual learning faces with temporal aspect of training as it deals with continuous streams of data. Distribution shifts over space and time is what it happens in real federated learning scenarios that show multiple challenges. First, the federated model needs to learn sequentially while retaining knowledge from the past training rounds. Second, the model has also to deal with concept drift from the distributed data distributions. To address these complexities, we attempt to combine continual and federated learning strategies by proposing a solution inspired by experience replay and generative adversarial concepts for supporting decentralized distributed training. In particular, our approach relies on using limited memory buffers of synthetic privacy-preserving samples and interleaving training on local data and on buffer data. By translating the CL formulation into the task of integrating distributed knowledge with local knowledge, our method enables models to effectively integrate learned representation from local nodes, providing models the capability to generalize across multiple datasets.We test our integrated strategy on two realistic medical image analysis tasks --- tuberculosis and melanoma classification --- using multiple datasets in order to simulate realistic non-i.i.d. medical data scenarios. Results show that our approach achieves performance comparable to standard (non-federated) learning and significantly outperforms state-of-the-art federated methods in their centralized (thus, more favourable) formulation.},
author = {Pennisi, Matteo and Proietto Salanitri, Federica and Bellitto, Giovanni and Casella, Bruno and Aldinucci, Marco and Palazzo, Simone and Spampinato, Concetto},
booktitle = {Proceedings of the 1st Workshop on Visual Continual Learning, ICCV 2023, Paris, France, 2 October 2023},
date-modified = {2024-04-01 11:13:18 +0200},
doi = {10.1109/ICCVW60793.2023.00362},
keywords = {ai},
note = {https://ieeexplore.ieee.org/document/10350429},
title = {Experience Replay as an Effective Strategy for Optimizing Decentralized Federated Learning},
url = {https://openaccess.thecvf.com/content/ICCV2023W/VCL/papers/Pennisi_Experience_Replay_as_an_Effective_Strategy_for_Optimizing_Decentralized_Federated_ICCVW_2023_paper.pdf},
year = {2023},
bdsk-url-1 = {https://openaccess.thecvf.com/content/ICCV2023W/VCL/papers/Pennisi_Experience_Replay_as_an_Effective_Strategy_for_Optimizing_Decentralized_Federated_ICCVW_2023_paper.pdf},
bdsk-url-2 = {https://doi.org/10.1109/ICCVW60793.2023.00362}}
Bruno Casella, Walter Riviera, Marco Aldinucci, Gloria Menegaz
Patterns
ai epi icsc
Driven by the deep learning (DL) revolution, artificial intelligence (AI) has become a fundamental tool for many biomedical tasks, including analyzing and classifying diagnostic images. Imaging, however, is not the only source of information. Tabular data, such as personal and genomic data and blood test results, are routinely collected but rarely considered in DL pipelines. Nevertheless, DL requires large datasets that often must be pooled from different institutions, raising non-trivial privacy concerns. Federated learning (FL) is a cooperative learning paradigm that aims to address these issues by moving models instead of data across different institutions. Here, we present a federated multi-input architecture using images and tabular data as a methodology to enhance model performance while preserving data privacy. We evaluated it on two showcases: the prognosis of COVID-19 and patients' stratification in Alzheimer's disease, providing evidence of enhanced accuracy and F1 scores against single-input models and improved generalizability against non-federated models.
@article{23:fl:patterns,
abstract = {Driven by the deep learning (DL) revolution, artificial intelligence (AI) has become a fundamental tool for many biomedical tasks, including analyzing and classifying diagnostic images. Imaging, however, is not the only source of information. Tabular data, such as personal and genomic data and blood test results, are routinely collected but rarely considered in DL pipelines. Nevertheless, DL requires large datasets that often must be pooled from different institutions, raising non-trivial privacy concerns. Federated learning (FL) is a cooperative learning paradigm that aims to address these issues by moving models instead of data across different institutions. Here, we present a federated multi-input architecture using images and tabular data as a methodology to enhance model performance while preserving data privacy. We evaluated it on two showcases: the prognosis of COVID-19 and patients' stratification in Alzheimer's disease, providing evidence of enhanced accuracy and F1 scores against single-input models and improved generalizability against non-federated models.},
author = {Bruno Casella and Walter Riviera and Marco Aldinucci and Gloria Menegaz},
date-modified = {2024-04-01 11:13:40 +0200},
doi = {10.1016/j.patter.2023.100856},
issn = {2666-3899},
journal = {Patterns},
keywords = {icsc, epi, ai},
pages = {100856},
title = {MERGE: A model for multi-input biomedical federated learning},
url = {https://www.sciencedirect.com/science/article/pii/S2666389923002404},
year = {2023},
bdsk-url-1 = {https://www.sciencedirect.com/science/article/pii/S2666389923002404},
bdsk-url-2 = {https://doi.org/10.1016/j.patter.2023.100856}}
Ovidio De Filippo, Francesco Bruno, Tineke H. Pinxterhuis, Mariusz Gasior, Leor Perl, Luca Gaido, Domenico Tuttolomondo, Antonio Greco, Roberto Verardi, Gianluca Lo Martire, Mario Iannaccone, Attilio Leone, Gaetano Liccardo, Serena Caglioni, Rocio González Ferreiro, Giulio Rodinò, Giuseppe Musumeci, Giuseppe Patti, Irene Borzillo, Giuseppe Tarantini, Wojciech Wańha, Bruno Casella, Eline H Ploumen, Lukasz Pyka, Ran Kornowski, Andrea Gagnor, Raffaele Piccolo, Sergio Raposeiras Roubin, Davide Capodanno, Paolo Zocca, Federico Conrotto, Gaetano Maria Ferrari, Clemens Birgelen, Fabrizio D'Ascenzo
Catheterization and Cardiovascular Interventions
ai cardio
Background: Data about the long-term performance of new-generation ultrathin-strut drug-eluting stents (DES) in challenging coronary lesions, such as left main (LM), bifurcation, and chronic total occlusion (CTO) lesions are scant. Methods: The international multicenter retrospective observational ULTRA study included consecutive patients treated from September 2016 to August 2021 with ultrathin-strut DES in challenging de novo lesions. Primary endpoint was target lesion failure (TLF): composite of cardiac death, target-lesion revascularization (TLR), target-vessel myocardial infarction (TVMI), or definite stent thrombosis (ST). Secondary endpoints included all-cause death, acute myocardial infarction (AMI), target vessel revascularization, and TLF components. TLF predictors were assessed with Cox multivariable analysis. Results: Of 1801 patients (age: 66.6±11.2 years; male: 1410 [78.3%]), 170 (9.4%) experienced TLF during follow-up of 3.1±1.4 years. In patients with LM, CTO, and bifurcation lesions, TLF rates were 13.5%, 9.9%, and 8.9%, respectively. Overall, 160 (8.9%) patients died (74 [4.1%] from cardiac causes). AMI and TVMI rates were 6.0% and 3.2%, respectively. ST occurred in 11 (1.1%) patients while 77 (4.3%) underwent TLR. Multivariable analysis identified the following predictors of TLF: age, STEMI with cardiogenic shock, impaired left ventricular ejection fraction, diabetes, and renal dysfunction. Among the procedural variables, total stent length increased TLF risk (HR: 1.01, 95% CI: 1-1.02 per mm increase), while intracoronary imaging reduced the risk substantially (HR: 0.35, 95% CI: 0.12-0.82). Conclusions: Ultrathin-strut DES showed high efficacy and satisfactory safety, even in patients with challenging coronary lesions. Yet, despite using contemporary gold-standard DES, the association persisted between established patient- and procedure-related features of risk and impaired 3-year clinical outcome.
@article{23:casella:ultra,
abstract = {Background: Data about the long-term performance of new-generation ultrathin-strut drug-eluting stents (DES) in challenging coronary lesions, such as left main (LM), bifurcation, and chronic total occlusion (CTO) lesions are scant. Methods: The international multicenter retrospective observational ULTRA study included consecutive patients treated from September 2016 to August 2021 with ultrathin-strut DES in challenging de novo lesions. Primary endpoint was target lesion failure (TLF): composite of cardiac death, target-lesion revascularization (TLR), target-vessel myocardial infarction (TVMI), or definite stent thrombosis (ST). Secondary endpoints included all-cause death, acute myocardial infarction (AMI), target vessel revascularization, and TLF components. TLF predictors were assessed with Cox multivariable analysis. Results: Of 1801 patients (age: 66.6$\pm$11.2 years; male: 1410 [78.3\%]), 170 (9.4\%) experienced TLF during follow-up of 3.1$\pm$1.4 years. In patients with LM, CTO, and bifurcation lesions, TLF rates were 13.5\%, 9.9\%, and 8.9\%, respectively. Overall, 160 (8.9\%) patients died (74 [4.1\%] from cardiac causes). AMI and TVMI rates were 6.0\% and 3.2\%, respectively. ST occurred in 11 (1.1\%) patients while 77 (4.3\%) underwent TLR. Multivariable analysis identified the following predictors of TLF: age, STEMI with cardiogenic shock, impaired left ventricular ejection fraction, diabetes, and renal dysfunction. Among the procedural variables, total stent length increased TLF risk (HR: 1.01, 95\% CI: 1-1.02 per mm increase), while intracoronary imaging reduced the risk substantially (HR: 0.35, 95\% CI: 0.12-0.82). Conclusions: Ultrathin-strut DES showed high efficacy and satisfactory safety, even in patients with challenging coronary lesions. Yet, despite using contemporary gold-standard DES, the association persisted between established patient- and procedure-related features of risk and impaired 3-year clinical outcome.},
author = {De Filippo, Ovidio and Bruno, Francesco and Pinxterhuis, Tineke H. and Gasior, Mariusz and Perl, Leor and Gaido, Luca and Tuttolomondo, Domenico and Greco, Antonio and Verardi, Roberto and Lo Martire, Gianluca and Iannaccone, Mario and Leone, Attilio and Liccardo, Gaetano and Caglioni, Serena and Gonz{\'a}lez Ferreiro, Rocio and Rodin{\`o}, Giulio and Musumeci, Giuseppe and Patti, Giuseppe and Borzillo, Irene and Tarantini, Giuseppe and Wa{\'n}ha, Wojciech and Casella, Bruno and Ploumen, Eline H and Pyka, Lukasz and Kornowski, Ran and Gagnor, Andrea and Piccolo, Raffaele and Roubin, Sergio Raposeiras and Capodanno, Davide and Zocca, Paolo and Conrotto, Federico and de Ferrari, Gaetano Maria and von Birgelen, Clemens and D'Ascenzo, Fabrizio},
date-modified = {2025-05-24 10:56:27 +0200},
doi = {10.1002/ccd.30696},
journal = {Catheterization and Cardiovascular Interventions},
keywords = {ai, cardio},
title = {Predictors of target lesion failure after treatment of left main, bifurcation, or chronic total occlusion lesions with ultrathin-strut drug-eluting coronary stents in the ULTRA registry},
url = {https://onlinelibrary.wiley.com/doi/full/10.1002/ccd.30696},
year = {2023},
bdsk-url-1 = {https://onlinelibrary.wiley.com/doi/full/10.1002/ccd.30696},
bdsk-url-2 = {https://doi.org/10.1002/ccd.30696}}
Iacopo Colonnelli, Bruno Casella, Gianluca Mittone, Yasir Arfat, Barbara Cantalupo, Roberto Esposito, Alberto R. Martinelli, Doriana Medić, Marco Aldinucci
Astrophysics and Space Science Proceedings
across eupilot fl streamflow
HPC and AI are fated to meet for several reasons. This article will discuss some of them and argue why this will happen through the set of methods and technologies that underpin cloud computing. As a paradigmatic example, we present a new federated learning system that collaboratively trains a deep learning model in different supercomputing centers. The system is based on the StreamFlow workflow manager designed for hybrid cloud-HPC infrastructures.
@inproceedings{22:ml4astro,
abstract = {HPC and AI are fated to meet for several reasons. This article will discuss some of them and argue why this will happen through the set of methods and technologies that underpin cloud computing. As a paradigmatic example, we present a new federated learning system that collaboratively trains a deep learning model in different supercomputing centers. The system is based on the StreamFlow workflow manager designed for hybrid cloud-HPC infrastructures.},
address = {Catania, Italy},
author = {Colonnelli, Iacopo and Casella, Bruno and Mittone, Gianluca and Arfat, Yasir and Cantalupo, Barbara and Esposito, Roberto and Martinelli, Alberto R. and Medi\'{c}, Doriana and Aldinucci, Marco},
booktitle = {Astrophysics and Space Science Proceedings},
date-modified = {2024-04-01 11:12:57 +0200},
doi = {10.1007/978-3-031-34167-0_39},
editor = {Bufano, Filomena and Riggi, Simone and Sciacca, Eva and Schillir\`{o}, Francesco},
isbn = {978-3-031-34167-0},
keywords = {across, eupilot, streamflow, fl},
pages = {193--199},
publisher = {Springer},
title = {Federated Learning meets {HPC} and cloud},
url = {https://iris.unito.it/retrieve/3ac66baa-9d9a-4e9f-94a5-13700694d8aa/ML4Astro.pdf},
volume = {60},
year = {2023},
bdsk-url-1 = {https://iris.unito.it/retrieve/3ac66baa-9d9a-4e9f-94a5-13700694d8aa/ML4Astro.pdf}}
Bruno Casella, Roberto Esposito, Carlo Cavazzoni, Marco Aldinucci
Proceedings of the 1st Italian Conference on Big Data and Data Science, ITADATA 2022, September 20-21, 2022
eupilot
Classic Machine Learning (ML) techniques require training on data available in a single data lake (either centralized or distributed). However, aggregating data from different owners is not always convenient for different reasons, including security, privacy and secrecy. Data carry a value that might vanish when shared with others; the ability to avoid sharing the data enables industrial applications where security and privacy are of paramount importance, making it possible to train global models by implementing only local policies which can be run independently and even on air-gapped data centres. Federated Learning (FL) is a distributed machine learning approach which has emerged as an effective way to address privacy concerns by only sharing local AI models while keeping the data decentralized. Two critical challenges of Federated Learning are managing the heterogeneous systems in the same federated network and dealing with real data, which are often not independently and identically distributed (non-IID) among the clients. In this paper, we focus on the second problem, i.e., the problem of statistical heterogeneity of the data in the same federated network. In this setting, local models might be strayed far from the local optimum of the complete dataset, thus possibly hindering the convergence of the federated model. Several Federated Learning algorithms, such as FedAvg, FedProx and Federated Curvature (FedCurv), aiming at tackling the non-IID setting, have already been proposed. This work provides an empirical assessment of the behaviour of FedAvg and FedCurv in common non-IID scenarios. Results show that the number of epochs per round is an important hyper-parameter that, when tuned appropriately, can lead to significant performance gains while reducing the communication cost. As a side product of this work, we release the non-IID version of the datasets we used so to facilitate further comparisons from the FL community.
@inproceedings{casella2022benchmarking,
abstract = {Classic Machine Learning (ML) techniques require training on data available in a single data lake (either centralized or distributed). However, aggregating data from different owners is not always convenient for different reasons, including security, privacy and secrecy. Data carry a value that might vanish when shared with others; the ability to avoid sharing the data enables industrial applications where security and privacy are of paramount importance, making it possible to train global models by implementing only local policies which can be run independently and even on air-gapped data centres. Federated Learning (FL) is a distributed machine learning approach which has emerged as an effective way to address privacy concerns by only sharing local AI models while keeping the data decentralized. Two critical challenges of Federated Learning are managing the heterogeneous systems in the same federated network and dealing with real data, which are often not independently and identically distributed (non-IID) among the clients. In this paper, we focus on the second problem, i.e., the problem of statistical heterogeneity of the data in the same federated network. In this setting, local models might be strayed far from the local optimum of the complete dataset, thus possibly hindering the convergence of the federated model. Several Federated Learning algorithms, such as FedAvg, FedProx and Federated Curvature (FedCurv), aiming at tackling the non-IID setting, have already been proposed. This work provides an empirical assessment of the behaviour of FedAvg and FedCurv in common non-IID scenarios. Results show that the number of epochs per round is an important hyper-parameter that, when tuned appropriately, can lead to significant performance gains while reducing the communication cost. As a side product of this work, we release the non-IID version of the datasets we used so to facilitate further comparisons from the FL community.},
author = {Bruno Casella and Roberto Esposito and Carlo Cavazzoni and Marco Aldinucci},
booktitle = {Proceedings of the 1st Italian Conference on Big Data and Data Science, {ITADATA} 2022, September 20-21, 2022},
date-modified = {2024-04-01 11:14:22 +0200},
editor = {Anisetti, Marco and Bonifati, Angela and Bena, Nicola and Ardagna, Claudio A. and Malerba, Donato},
keywords = {eupilot},
publisher = {CEUR-WS.org},
series = {{CEUR} Workshop Proceedings},
title = {Benchmarking FedAvg and FedCurv for Image Classification Tasks},
url = {https://ceur-ws.org/Vol-3340/paper40.pdf},
volume = {3340},
year = {2022},
bdsk-url-1 = {https://iris.unito.it/bitstream/2318/1870961/1/Benchmarking_FedAvg_and_FedCurv_for_Image_Classification_Tasks.pdf}}
Bruno Casella, Alessio Chisari, Sebastiano Battiato, Mario Giuffrida
Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2022, Volume 5: VISAPP, Online Streaming, February 6-8, 2022
ai
It has been demonstrated that deep neural networks outperform traditional machine learning. However, deep networks lack generalisability, that is, they will not perform as good as in a new (testing) set drawn from a different distribution due to the domain shift. In order to tackle this known issue, several transfer learning approaches have been proposed, where the knowledge of a trained model is transferred into another to improve performance with different data. However, most of these approaches require additional training steps, or they suffer from catastrophic forgetting that occurs when a trained model has overwritten previously learnt knowledge. We address both problems with a novel transfer learning approach that uses network aggregation. We train dataset-specific networks together with an aggregation network in a unified framework. The loss function includes two main components: a task-specific loss (such as cross-entropy) and an aggregation loss. The proposed aggregation loss allows our model to learn how trained deep network parameters can be aggregated with an aggregation operator. We demonstrate that the proposed approach learns model aggregation at test time without any further training step, reducing the burden of transfer learning to a simple arithmetical operation. The proposed approach achieves comparable performance w.r.t. the baseline. Besides, if the aggregation operator has an inverse, we will show that our model also inherently allows for selective forgetting, i.e., the aggregated model can forget one of the datasets it was trained on, retaining information on the others.
@inproceedings{22:VISAPP:transferlearning,
abstract = {It has been demonstrated that deep neural networks outperform traditional machine learning. However, deep networks lack generalisability, that is, they will not perform as good as in a new (testing) set drawn from a different distribution due to the domain shift. In order to tackle this known issue, several transfer learning approaches have been proposed, where the knowledge of a trained model is transferred into another to improve performance with different data. However, most of these approaches require additional training steps, or they suffer from catastrophic forgetting that occurs when a trained model has overwritten previously learnt knowledge. We address both problems with a novel transfer learning approach that uses network aggregation. We train dataset-specific networks together with an aggregation network in a unified framework. The loss function includes two main components: a task-specific loss (such as cross-entropy) and an aggregation loss. The proposed aggregation loss allows our model to learn how trained deep network parameters can be aggregated with an aggregation operator. We demonstrate that the proposed approach learns model aggregation at test time without any further training step, reducing the burden of transfer learning to a simple arithmetical operation. The proposed approach achieves comparable performance w.r.t. the baseline. Besides, if the aggregation operator has an inverse, we will show that our model also inherently allows for selective forgetting, i.e., the aggregated model can forget one of the datasets it was trained on, retaining information on the others.},
author = {Bruno Casella and Alessio Chisari and Sebastiano Battiato and Mario Giuffrida},
booktitle = {Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, {VISIGRAPP} 2022, Volume 5: VISAPP, Online Streaming, February 6-8, 2022},
date-modified = {2024-04-01 11:13:11 +0200},
doi = {10.5220/0010907900003124},
editor = {Giovanni Maria Farinella and Petia Radeva and Kadi Bouatouch},
isbn = {978-989-758-555-5},
keywords = {ai},
organization = {INSTICC},
pages = {642--649},
publisher = {SciTePress},
title = {Transfer Learning via Test-time Neural Networks Aggregation},
url = {https://iris.unito.it/retrieve/handle/2318/1844159/947123/TRANSFER_LEARNING_VIA_TEST_TIME_NEURAL_NETWORKS_AGGREGATION.pdf},
year = {2022},
bdsk-url-1 = {https://iris.unito.it/retrieve/handle/2318/1844159/947123/TRANSFER_LEARNING_VIA_TEST_TIME_NEURAL_NETWORKS_AGGREGATION.pdf},
bdsk-url-2 = {https://doi.org/10.5220/0010907900003124}}
Federated Learning (FL) is a distributed machine learning approach that has emerged as an effective way to address recent privacy concerns. However, FL introduces the need for additional security measures as FL alone is still subject to vulnerabilities such as model and data poisoning and inference attacks. Confidential Computing (CC) is a paradigm that, by leveraging hardware-based trusted execution environments (TEEs), protects the confidentiality and integrity of ML models and data, thus resulting in a powerful ally of FL applications. Typical TEEs offer an application-isolation level but suffer many drawbacks, such as limited available memory and debugging and coding difficulties. The new generation of TEEs offers a virtual machine (VM)-based isolation level, thus reducing the porting effort for existing applications. In this work, we compare the performance of VM-based and application-isolation level TEEs for confidential FL (CFL) applications. In particular, we evaluate the impact of TEEs and additional security mechanisms such as TLS (for securing the communication channel). The results, obtained across three datasets and two deep learning models, demonstrate that the new VM-based TEEs introduce a limited overhead (at most 1.5x), thus paving the way to leverage public and untrusted computing environments, such as HPC facilities or public cloud, without detriment to performance.
@misc{25:casella:TDXPDP,
abstract = {Federated Learning (FL) is a distributed machine learning approach that has emerged as an effective way to address recent privacy concerns. However, FL introduces the need for additional security measures as FL alone is still subject to vulnerabilities such as model and data poisoning and inference attacks. Confidential Computing (CC) is a paradigm that, by leveraging hardware-based trusted execution environments (TEEs), protects the confidentiality and integrity of ML models and data, thus resulting in a powerful ally of FL applications. Typical TEEs offer an application-isolation level but suffer many drawbacks, such as limited available memory and debugging and coding difficulties. The new generation of TEEs offers a virtual machine (VM)-based isolation level, thus reducing the porting effort for existing applications. In this work, we compare the performance of VM-based and application-isolation level TEEs for confidential FL (CFL) applications. In particular, we evaluate the impact of TEEs and additional security mechanisms such as TLS (for securing the communication channel). The results, obtained across three datasets and two deep learning models, demonstrate that the new VM-based TEEs introduce a limited overhead (at most 1.5x), thus paving the way to leverage public and untrusted computing environments, such as HPC facilities or public cloud, without detriment to performance.},
address = {Torino, Italy},
author = {Bruno Casella},
howpublished = {33rd Euromicro International Conference on Parallel, Distributed and Network-based Processing (PDP 2025)},
keywords = {icsc},
month = mar,
title = {A Performance Analysis of VM-based Trusted Execution Environments for Confidential Federated Learning},
url = {https://datacloud.di.unito.it/index.php/s/CFjA4sqYkbw9Bjy},
year = 2025,
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/CFjA4sqYkbw9Bjy}}
Bruno Casella, Iacopo Colonnelli, Gianluca Mittone, Robert Birke, Walter Riviera, Antonio Sciarappa, Carlo Cavazzoni, Marco Aldinucci
icsc
Federated Learning (FL) has emerged as a solution to preserve data privacy by keeping the data locally on each participant's device. However, FL alone is still vulnerable to attacks that can cause privacy leaks. Therefore, additional security measures, at the cost of increasing runtimes, become necessary. The Trusted Execution Environment (TEE) approach offers the highest degree of security during execution. However, TEEs suffer from memory limits which prevent safe end-to-end FL training of modern deep models. State-of-the-art approaches limit secure training to selected layers, failing to avert the full spectrum of attacks or adopt layer-wise training affecting model performance. We benchmark the usage of a library OS (LibOS) to run the full, unmodified end-to-end FL training inside the TEE. We extensively evaluate and model the overhead of the different security mechanisms needed to protect the data and model during computation (TEE), communication (TLS), and storage (disk encryption). The obtained results across three datasets and two models demonstrate that LibOSes are a viable way to seamlessly inject security into FL with limited overhead (at most 2x), offering valuable guidance for researchers and developers aiming to apply FL in data-security-focused contexts.
@misc{25:casella:sgx_sfo,
abstract = {Federated Learning (FL) has emerged as a solution to preserve data privacy by keeping the data locally on each participant's device. However, FL alone is still vulnerable to attacks that can cause privacy leaks. Therefore, additional security measures, at the cost of increasing runtimes, become necessary. The Trusted Execution Environment (TEE) approach offers the highest degree of security during execution. However, TEEs suffer from memory limits which prevent safe end-to-end FL training of modern deep models. State-of-the-art approaches limit secure training to selected layers, failing to avert the full spectrum of attacks or adopt layer-wise training affecting model performance. We benchmark the usage of a library OS (LibOS) to run the full, unmodified end-to-end FL training inside the TEE. We extensively evaluate and model the overhead of the different security mechanisms needed to protect the data and model during computation (TEE), communication (TLS), and storage (disk encryption). The obtained results across three datasets and two models demonstrate that LibOSes are a viable way to seamlessly inject security into FL with limited overhead (at most 2x), offering valuable guidance for researchers and developers aiming to apply FL in data-security-focused contexts.},
address = {San Francisco, US},
author = {Bruno Casella and Iacopo Colonnelli and Gianluca Mittone and Robert Birke and Walter Riviera and Antonio Sciarappa and Carlo Cavazzoni and Marco Aldinucci},
howpublished = {IEEE Security and Privacy Workshops (SPW)},
keywords = {icsc},
month = may,
title = {A Performance Analysis for Confidential Federated Learning},
url = {https://datacloud.di.unito.it/index.php/s/pzx75FGgP43YybF},
year = 2025,
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/pzx75FGgP43YybF}}
Bruno Casella, Matthias Jakobs, Marco Aldinucci, Sebastian Buschjaeger
icsc
Time series classification (TSC) is a critical task with applications in various domains, including healthcare, finance, and industrial monitoring. Due to privacy concerns and data regulations, Federated Learning has emerged as a promising approach for learning from distributed time series data without centralizing raw information. However, most FL solutions rely on a client-server architecture, which introduces robustness and confidentiality risks related to the distinguished role of the server, which is a single point of failure and can observe knowledge extracted from clients. To address these challenges, we propose DROCKS, a fully decentralized FL framework for TSC that leverages ROCKET (RandOm Convolutional KErnel Transform) features. In DROCKS, the global model is trained by sequentially traversing a structured path across federation nodes, where each node refines the model and selects the most effective local kernels before passing them to the successor. Extensive experiments on the UCR archive demonstrate that DROCKS outperforms state-of-the-art client-server FL approaches while being more resilient to node failures and malicious attacks.
@misc{25:casella:drocks,
abstract = {Time series classification (TSC) is a critical task with applications in various domains, including healthcare, finance, and industrial monitoring. Due to privacy concerns and data regulations, Federated Learning has emerged as a promising approach for learning from distributed time series data without centralizing raw information. However, most FL solutions rely on a client-server architecture, which introduces robustness and confidentiality risks related to the distinguished role of the server, which is a single point of failure and can observe knowledge extracted from clients. To address these challenges, we propose DROCKS, a fully decentralized FL framework for TSC that leverages ROCKET (RandOm Convolutional KErnel Transform) features. In DROCKS, the global model is trained by sequentially traversing a structured path across federation nodes, where each node refines the model and selects the most effective local kernels before passing them to the successor. Extensive experiments on the UCR archive demonstrate that DROCKS outperforms state-of-the-art client-server FL approaches while being more resilient to node failures and malicious attacks.},
address = {Porto, Portugal},
author = {Bruno Casella and Matthias Jakobs and Marco Aldinucci and Sebastian Buschjaeger},
howpublished = {3rD Workshop on Advancements in Federated Learning (ECML-PKDD)},
keywords = {icsc},
month = sep,
title = {Decentralized Time Series Classification with ROCKET features},
url = {https://datacloud.di.unito.it/index.php/s/Bp7Y5NmDDBwcizH},
year = 2025,
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/Bp7Y5NmDDBwcizH}}
Bruno Casella, Samuele Fonio, Lorenzo Sciandra, Claudio Gallicchio, Marco Aldinucci, Mirko Polato, Roberto Esposito
icsc
Time series classification is a pivotal task in modern machine learning, with widespread applications in fields such as healthcare, finance, and cybersecurity. While deep learning methods dominate recent developments, their resource demands and privacy limitations hinder deployment on low-power and decentralized environments. To address these challenges, we introduce Fed2RC, a fully federated and gradient-free approach that integrates the efficiency of Rocket-based feature extraction with the robustness of ridge regression in a privacy-preserving setting. Fed2RC builds upon two key ideas: (i) federated selection and aggregation of high-performing random convolution kernels, and (ii) incremental and communication-efficient updates of ridge classifier parameters using closed-form solutions. Additionally, we propose a novel federated protocol for selecting the global ridge regularization parameter lambda, and show how to improve the communication efficiency by matrix factorization techniques. Extensive experiments on the UCR benchmark demonstrate that Fed2RC achieves state-of-the-art results with a fraction of the computation and communication costs. Code to reproduce the experiments can be found at: https://github.com/CasellaJr/Fed2RC.
@misc{25:casella:Fed2RC,
abstract = {Time series classification is a pivotal task in modern machine learning, with widespread applications in fields such as healthcare, finance, and cybersecurity. While deep learning methods dominate recent developments, their resource demands and privacy limitations hinder deployment on low-power and decentralized environments. To address these challenges, we introduce Fed2RC, a fully federated and gradient-free approach that integrates the efficiency of Rocket-based feature extraction with the robustness of ridge regression in a privacy-preserving setting. Fed2RC builds upon two key ideas: (i) federated selection and aggregation of high-performing random convolution kernels, and (ii) incremental and communication-efficient updates of ridge classifier parameters using closed-form solutions. Additionally, we propose a novel federated protocol for selecting the global ridge regularization parameter lambda, and show how to improve the communication efficiency by matrix factorization techniques. Extensive experiments on the UCR benchmark demonstrate that Fed2RC achieves state-of-the-art results with a fraction of the computation and communication costs. Code to reproduce the experiments can be found at: https://github.com/CasellaJr/Fed2RC.},
address = {Bologna, Italy},
author = {Bruno Casella and Samuele Fonio and Lorenzo Sciandra and Claudio Gallicchio and Marco Aldinucci and Mirko Polato and Roberto Esposito},
howpublished = {28th European Conference on Artificial Intelligence},
keywords = {icsc},
month = oct,
title = {Fed2RC: Federated Rocket Kernels and Ridge Classifier for Time Series Classification},
url = {https://datacloud.di.unito.it/index.php/s/CKpYNTrZ3KDgkc6},
year = 2025,
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/CKpYNTrZ3KDgkc6}}
Marco Aldinucci, Gianluca Mittone, Bruno Casella, Robert Birke
fl hpc4ai icsc
@misc{25:huawei:fl,
address = {Zurich, Switzerland},
author = {Marco Aldinucci and Gianluca Mittone and Bruno Casella and Robert Birke},
howpublished = {Huawei Snow Mountain 2025},
keywords = {icsc, hpc4ai, fl},
month = nov,
title = {Decentralized Machine Learning and the Compute Continuum},
year = 2025}
Oussama Harrak, Bruno Casella, Samuele Fonio, Piero Fariselli, Gianluca Mittone, Cesare Rollo, Tiziana Sanavia, Marco Aldinucci
ai epi fl icsc
This work proposes FedSurvBoost, a federated learning pipeline for survival analysis based on the AdaBoost.F algorithm, which iteratively aggregates the best local weak hypotheses. Our method extends AdaBoost.F by removing the dependence on the number of classes coefficient from the computation of the weights of the best model. This makes it suitable for regression tasks, such as survival analysis. We show the effectiveness of our approach by comparing it with state-of-the-art methods, specifically developed for survival analysis problems, on two common survival datasets.
@misc{23:ecmlpkdd:fedsurvboost,
abstract = {This work proposes FedSurvBoost, a federated learning pipeline for survival analysis based on the AdaBoost.F algorithm, which iteratively aggregates the best local weak hypotheses. Our method extends AdaBoost.F by removing the dependence on the number of classes coefficient from the computation of the weights of the best model. This makes it suitable for regression tasks, such as survival analysis. We show the effectiveness of our approach by comparing it with state-of-the-art methods, specifically developed for survival analysis problems, on two common survival datasets.},
address = {Vilnius, Lithuania},
author = {Oussama Harrak and Bruno Casella and Samuele Fonio and Piero Fariselli and Gianluca Mittone and Cesare Rollo and Tiziana Sanavia and Marco Aldinucci},
howpublished = {European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2nd Workshop on Advancements in Federated Learning (WAFL)},
keywords = {icsc, epi, ai, fl},
month = sep,
title = {Federated Adaboost for Survival Analysis},
url = {https://datacloud.di.unito.it/index.php/s/DtXiQfne6BEC235},
year = 2024,
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/DtXiQfne6BEC235}}
Bruno Casella, Matthias Jakobs, Marco Aldinucci, Sebastian Buschjaeger
icsc
This paper proposes FROCKS, a federated time series classification method using ROCKET features. Our approach dynamically adapts the models' features by selecting and exchanging the best-performing ROCKET kernels from a federation of clients. Specifically, the server gathers the best-performing kernels of the clients together with the associated model parameters, and it performs a weighted average if a kernel is best-performing for more than one client. We compare the proposed method with state-of-the-art approaches on the UCR archive binary classification datasets and show superior performance on most datasets.
@misc{25:casella:frocks,
abstract = {This paper proposes FROCKS, a federated time series classification method using ROCKET features. Our approach dynamically adapts the models' features by selecting and exchanging the best-performing ROCKET kernels from a federation of clients. Specifically, the server gathers the best-performing kernels of the clients together with the associated model parameters, and it performs a weighted average if a kernel is best-performing for more than one client. We compare the proposed method with state-of-the-art approaches on the UCR archive binary classification datasets and show superior performance on most datasets.},
address = {Bruges, Belgium},
author = {Bruno Casella and Matthias Jakobs and Marco Aldinucci and Sebastian Buschjaeger},
howpublished = {32nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning},
keywords = {icsc},
month = oct,
title = {Federated Time Series Classification with ROCKET features},
url = {https://datacloud.di.unito.it/index.php/s/xWCBsMfjZMkyrAq},
year = 2024,
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/xWCBsMfjZMkyrAq}}
Proceedings of the 3rd Workshop on Distributed Machine Learning for the Intelligent Computing Continuum (DML-ICC), IEEE/ACM UCC 2023, Taormina, Italy, 4 December 2023
ai epi fl icsc
Federated Learning (FL) has emerged as a promising solution to address privacy concerns by collaboratively training Deep Learning (DL) models across distributed parties. This work proposes an architecture-based aggregation strategy in Vertical FL, where parties hold data with different attributes but shared instances. Our approach leverages the identical architectural parts, i.e. neural network layers, of different models to selectively aggregate weights, which is particularly relevant when collaborating with institutions holding different types of datasets, i.e., image, text, or tabular datasets. In a scenario where two entities train DL models, such as a Convolutional Neural Network (CNN) and a Multi-Layer Perceptron (MLP), our strategy computes the average only for architecturally identical segments. This preserves data-specific features learned from demographic and clinical data. We tested our approach on two clinical datasets, i.e., the COVID-CXR dataset and the ADNI study. Results show that our method achieves comparable results with the centralized scenario, in which all the data are collected in a single data lake, and benefits from FL generalizability. In particular, compared to the non-federated models, our proposed proof-of-concept model exhibits a slight performance loss on the COVID-CXR dataset (less than 8
@misc{23:casella:architecturalfedavgtalk,
abstract = {Federated Learning (FL) has emerged as a promising solution to address privacy concerns by collaboratively training Deep Learning (DL) models across distributed parties. This work proposes an architecture-based aggregation strategy in Vertical FL, where parties hold data with different attributes but shared instances. Our approach leverages the identical architectural parts, i.e. neural network layers, of different models to selectively aggregate weights, which is particularly relevant when collaborating with institutions holding different types of datasets, i.e., image, text, or tabular datasets. In a scenario where two entities train DL models, such as a Convolutional Neural Network (CNN) and a Multi-Layer Perceptron (MLP), our strategy computes the average only for architecturally identical segments. This preserves data-specific features learned from demographic and clinical data. We tested our approach on two clinical datasets, i.e., the COVID-CXR dataset and the ADNI study. Results show that our method achieves comparable results with the centralized scenario, in which all the data are collected in a single data lake, and benefits from FL generalizability. In particular, compared to the non-federated models, our proposed proof-of-concept model exhibits a slight performance loss on the COVID-CXR dataset (less than 8%), but outperforms ADNI models by up to 12%. Moreover, communication costs between training rounds are minimized by exchanging only the dense layer parameters.},
author = {Casella, Bruno and Fonio, Samuele},
booktitle = {Proceedings of the 3rd Workshop on Distributed Machine Learning for the Intelligent Computing Continuum (DML-ICC), IEEE/ACM UCC 2023, Taormina, Italy, 4 December 2023},
keywords = {icsc, epi, ai, fl},
title = {Architecture-Based FedAvg for Vertical Federated Learning},
url = {https://datacloud.di.unito.it/index.php/s/kJQxnqG4d2ZSicK},
year = 2023,
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/kJQxnqG4d2ZSicK}}
Presentation of the paper "Transfer Learning via Test-Time Neural Networks Aggregation" to VISAPP, the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications
@misc{22:visapp,
abstract = {Presentation of the paper "Transfer Learning via Test-Time Neural Networks Aggregation" to VISAPP, the 19th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications},
address = {Virtual event},
author = {Bruno Casella},
howpublished = {VISAPP},
month = feb,
title = {Transfer Learning via Test-Time Neural Networks Aggregation},
url = {https://datacloud.di.unito.it/index.php/s/AKMT87eZDFX46JJ},
year = 2022,
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/AKMT87eZDFX46JJ}}
Bruno Casella, Gianluca Mittone, Yasir Arfat, Marco Aldinucci, Roberto Esposito
ai fl
Since its debut in 2016, Federated Learning (FL) has been tied to the inner workings of Deep Neural Networks (DNNs). On the one hand, this allowed its development and widespread use as DNNs proliferated. On the other hand, it neglected all those scenarios in which using DNNs is not possible or advantageous. The fact that most current FL frameworks only allow training DNNs reinforces this problem. To address the lack of FL solutions for non-DNN-based use cases, we propose MAFL (Model-Agnostic Federated Learning). MAFL marries a model-agnostic FL algorithm, AdaBoost.F, with an open industry-grade FL framework: Intel® OpenFL. MAFL is the first FL system not tied to any specific type of machine learning model, allowing exploration of FL scenarios beyond DNNs and trees. Furthermore, tools for DML (especially FL) are starting to flourish, many are not flexible and portable enough to experiment with novel systems (e.g., RISC-V), non-fully connected topologies, and asynchronous collaboration schemes. We overcome these limitations via a domain-specific language allowing to map DML schemes to an underlying middleware, i.e. the \ffparallel programming library.
@misc{22:ITWSHPC:FL,
abstract = {Since its debut in 2016, Federated Learning (FL) has been tied to the inner workings of Deep Neural Networks (DNNs). On the one hand, this allowed its development and widespread use as DNNs proliferated. On the other hand, it neglected all those scenarios in which using DNNs is not possible or advantageous. The fact that most current FL frameworks only allow training DNNs reinforces this problem. To address the lack of FL solutions for non-DNN-based use cases, we propose MAFL (Model-Agnostic Federated Learning). MAFL marries a model-agnostic FL algorithm, AdaBoost.F, with an open industry-grade FL framework: Intel{\textregistered} OpenFL. MAFL is the first FL system not tied to any specific type of machine learning model, allowing exploration of FL scenarios beyond DNNs and trees. Furthermore, tools for DML (especially FL) are starting to flourish, many are not flexible and portable enough to experiment with novel systems (e.g., RISC-V), non-fully connected topologies, and asynchronous collaboration schemes. We overcome these limitations via a domain-specific language allowing to map DML schemes to an underlying middleware, i.e. the \ff parallel programming library.},
address = {Turin, Italy},
annote = {invited},
author = {Bruno Casella and Gianluca Mittone and Yasir Arfat and Marco Aldinucci and Roberto Esposito},
date-modified = {2026-09-13 23:04:13 +0200},
howpublished = {3rd Italian Workshop on HPC},
keywords = {fl,ai},
month = sep,
note = {Invited talk},
title = {Towards Machine Learning on the edge: the opportunities of Federated Learning},
url = {https://datacloud.di.unito.it/index.php/s/X3bDLCBeZMo4qr5},
year = {2022},
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/X3bDLCBeZMo4qr5}}
Presentation of the paper "Benchmarking FedAvg and FedCurv for Image Classification Tasks" to the first italian conference on Big Data and Data Science
@misc{22:itadata,
abstract = {Presentation of the paper "Benchmarking FedAvg and FedCurv for Image Classification Tasks" to the first italian conference on Big Data and Data Science},
address = {Milan, Italy},
author = {Bruno Casella},
howpublished = {ITADATA},
keywords = {eupilot},
month = sep,
title = {Benchmarking FedAvg and FedCurv for Image Classification Tasks},
url = {https://datacloud.di.unito.it/index.php/s/6XaEXnAowRrAHGL},
year = {2022},
bdsk-url-1 = {https://datacloud.di.unito.it/index.php/s/6XaEXnAowRrAHGL}}
