Archivio della ricerca - Fondazione Bruno Kessler
Not a member yet
21227 research outputs found
Sort by
Replay Consolidation with Label Propagation for Continual Object Detection
No full textContinual Learning (CL) aims to learn new data while remembering previously acquired knowledge. In contrast to CL for image classification, CL for Object Detection faces additional challenges such as the missing annotations problem. In this scenario, images from previous tasks may contain instances of unknown classes that could reappear as labeled in future tasks, leading to task interference in replay-based approaches. Consequently, most approaches in the literature have focused on distillation-based techniques, which are effective when there is a significant class overlap between tasks. In our work, we propose an alternative to distillation-based approaches with a novel approach called Replay Consolidation with Label Propagation for Object Detection (RCLPOD). RCLPOD enhances the replay memory by improving the quality of the stored samples through a technique that promotes class balance while also improving the quality of the ground truth associated with these samples through a technique called label propagation. RCLPOD outperforms existing techniques on well-established benchmarks such as VOC and COC. Moreover, our approach is developed to work with modern architectures like YOLOv8, making it suitable for dynamic, real-world applications such as autonomous driving and robotics, where continuous learning and resource efficiency are essential
A Tool for the Implementation of Open Neural Network Exchange Models in Functional Mockup Units
Get PDFThe Functional Mock-up Interface (FMI) standard is aflagship in the co-simulation and model exchange domain.However, the integration of graph-based computationalmodels—particularly neural networks—into Functional Mock-upUnits (FMUs) has remained a technical challenge due tointeroperability and platform-specific limitations.To address this, we propose ONNX2FMU, a command-line Pythontool that facilitates the deployment of Open Neural NetworkExchange (ONNX) models into FMUs. According to FMI's goodpractices, ONNX2FMU generates C source code to wrap ONNXmodels in Functional Mockup Units, supports FMI versions2.0 and 3.0, and provides multi-platform compilationcapabilities. The tool simplifies the mapping processbetween model description and ONNX model inputs and outputsvia JSON files, ensuring accessibility and flexibility.This paper presents the tool architecture and methodologyand showcases its applicability through illustrativeexamples, including a reduced-order model powered by arecurrent neural network
ATLAS searches for additional scalars and exotic Higgs boson decays with the LHC Run 2 dataset
No full textThis report reviews the published results of searches for possible additional scalar particles and exotic decays of the Higgs boson performed by the ATLAS Collaboration using up to 140 fb−1 of 13 TeV proton–proton collision data collected during Run 2 of the Large Hadron Collider. Key results are examined, and observed excesses, while never statistically compelling, are noted. Constraints are placed on parameters of several models which extend the Standard Model, for example by adding one or more singlet or doublet fields, or offering exotic Higgs boson decay channels. Summaries of new searches as well as extensions of previous searches are discussed. These new results have a wider reach or attain stronger exclusion limits. New experimental techniques that were developed for these searches are highlighted. Search channels which have not yet been examined are also listed, as these provide insight into possible future areas of exploration
Mixed-Signal Silicon Photomultiplier With Reconfigurable Pulse Shaper for Background Light Rejection
No full textTracking Seasonal Influenza Trends in South Tyrol During 2022/2023 Using Genomic Surveillance Data
No full textBackground: Influenza seasons are characterized by a complex interplay of co-circulating strains with high spatial and temporal heterogeneities. Surveillance is crucial for monitoring the spread and evolution of the virus and design effective public health response strategies. Aim: We combined epidemiological, virological, and genomic surveillance data to provide a comprehensive analysis of influenza subtypes circulating in the South Tyrol region (Italy) during season 2022/2023, leveraging phylogenetic and phylodynamic approaches. Methods: Clinical samples were collected from patients exhibiting influenza-like symptoms and evaluated by molecular diagnostics. Whole genome sequencing was conducted, and the hemagglutinin (HA) gene sequences were used for phylogenetic analysis. A birth-death skyline model was applied to estimate strain-specific effective reproduction numbers (Re) and attack rates. Results: Out of 4891 samples tested, 862 tested positive for influenza, of which 224 genomes were sequenced. Phylogenetic analysis of HA gene revealed A(H3N2) strains predominantly clustering in clade 3C.2a1b.2a.2b, followed by 3C.2a1b.2a.1b. A(H1N1pdm09) strains predominantly clustered in clade 6B.1A.5a.2a. Exclusive circulation of B (Victoria) subtype strains aligned with the global trend, all falling within clade V1A.3a.2. Phylogenetic analyses indicate that the strains isolated in the South Tyrol region closely resembled those circulating in the rest of Italy and Austria. Re peaked at 1.16-1.35 (95%CI) for A(H3N2), 1.06-1.34 for A(H1N1pdm09) and 1.02-1.29 for B (Victoria). 95%CI of attack rates were 6.3%-33.5% for A(H3N2), 0.6%-5.0% for A(H1N1pdm09), and 0.8%-6.5% for B (Victoria). Conclusion: Epidemiological estimates from traditional surveillance data can be corroborated by those derived from genomic sequencing, providing more robust assessments of viral transmissibility and attack rates with limited additional effort
The third wheel or the game changer? How AI could team up with neurologists in Parkinson's care
No full textIntroduction: Parkinson's disease (PD) is a progressive neurodegenerative disorder marked by diverse motor and non-motor symptoms. AI appears to be the elephant in the neurologist's room, albeit offering transformative potential in early diagnosis, personalized care, and treatment optimization, addressing the complexities of PD management. Methods: A PubMed review analyzed AI's role in PD care, focusing on multimodal data, clinician-in-the-loop models, and validation strategies to enhance clinical applicability while addressing ethical concerns. Results: Thirty-nine studies met inclusion criteria. While progress is evident, challenges include limited independent validation, small sample sizes, and inconsistent clinician involvement. Key trends show growing interest in multimodal data and predictive modelling, though gaps in transparency persist. Conclusions: AI holds promise for improving PD management, yet robust validation, interdisciplinary collaboration, and clearer clinician roles are essential for advancing patient-centered care and fostering trust in AI-driven innovations
Runtime integration of machine learning and simulation for business processes: Time and decision mining predictions
No full textRecent research in Computer Science has investigated the use of Deep Learning (DL) techniques to complement outcomes or decisions within a Discrete Event Simulation (DES) model. The main idea of this combination is to maintain a white box simulation model complement it with information provided by DL models to overcome the unrealistic or oversimplified assumptions of traditional DESs. State-of-the-art techniques in BPM combine Deep Learning and Discrete Event Simulation in a post-integration fashion: first an entire simulation is performed, and then a DL model is used to add waiting times and processing times to the events produced by the simulation model.
In this paper, we aim at taking a step further by introducing Rims (Runtime Integration of Machine Learning and Simulation). Instead of complementing the outcome of a complete simulation with the results of predictions a posteriori, Rims provides a tight integration of the predictions of the DL model at runtime during the simulation. This runtime-integration enables us to fully exploit the specific predictions while respecting simulation execution, thus enhancing the performance of the overall system both w.r.t. the single techniques (Business Process Simulation and DL) separately and the post-integration approach. In particular, the runtime integration ensures the accuracy of intercase features for time prediction, such as the number of ongoing traces at a given time, by calculating them during directly the simulation, where all traces are executed in parallel. Additionally, it allows for the incorporation of online queue information in the DL model and enables the integration of other predictive models into the simulator to enhance decision point management within the process model. These enhancements improve the performance of Rims in accurately simulating the real process in terms of control flow, as well as in terms of time and congestion dimensions. Especially in process scenarios with significant congestion – when a limited availability of resources leads to significant event queues for their allocation – the ability of Rims to use queue features to predict waiting times allows it to surpass the state-of-the-art. We evaluated our approach with real-world and synthetic event logs, using various metrics to assess the simulation model’s quality in terms of control-flow, time, and congestion dimensions
Software and computing for Run 3 of the ATLAS experiment at the LHC
No full textThe ATLAS experiment has developed extensive software and distributed computing systems for Run 3 of the LHC. These systems are described in detail, including software infrastructure and workflows, distributed data and workload management, database infrastructure, and validation. The use of these systems to prepare the data for physics analysis and assess its quality are described, along with the software tools used for data analysis itself. An outlook for the development of these projects towards Run 4 is also provided