Archivio della ricerca - Fondazione Bruno Kessler
Not a member yet
21227 research outputs found
Sort by
Investigation of the surface properties of the LISA Pathfinder witness samples and their implications on the LISA discharge system
No full textWe report a series of measurements of the surface properties on gold-coated samples of the LISA Pathfinder (LPF) test mass and electrode housing surfaces [1], that are the core of the Gravitational Reference System (GRS) for the space based gravitational wave observatory LISA (Laser Interferometer Space Antenna). This work gives insight into preparing the analysis plan for similar samples that will be produced together with the GRS prototypes models for LISA in the next years [2]. This study aims at understanding the dynamic changes of the surface properties and consolidate a strategy to maintain and, eventually, recover electron photoemission. Understanding if and, eventually, how surface properties could change will help optimize the manufacturing process, storage environment, and bake-out temperature of the gold-coated surfaces
Granary: Speech Recognition and Translation Dataset in 25 European Languages
Get PDFMulti-task and multilingual approaches benefit large models, yet speech processing for low-resource languages remains underexplored due to data scarcity. To address this, we present Granary, a large-scale collection of speech datasets for recognition and translation across 25 European languages. This is the first open-source effort at this scale for both transcription and translation. We enhance data quality using a pseudo-labeling pipeline with segmentation, two-pass inference, hallucination filtering, and punctuation restoration. We further generate translation pairs from pseudo-labeled transcriptions using EuroLLM, followed by a data filtration pipeline. Designed for efficiency, our pipeline processes vast amount of data within hours. We assess models trained on processed data by comparing their performance on previously curated datasets for both high- and low-resource languages. Our findings show that these models achieve similar performance using approx. 50% less data. Dataset will be made available at https://hf.co/datasets/nvidia/Granary
Formation of luminescent defects in 4H-SiC upon ion irradiation and ns laser annealing
No full textWe present a novel approach for the optical activation of the negatively-charged silicon vacancy ( VSi−) center in ion irradiated silicon carbide (SiC) via ns-pulsed laser annealing in the 234–2180 mJ cm− 2 energy density range. The laser annealing process is investigated under 355 nm and 532 nm wavelengths at pulse energy densities below the melting threshold and validated by means of Raman spectroscopy and photoluminescence mapping. The combined effect of ns pulsed laser annealing and subsequent thermal treatment is also assessed. The results offer a promising resource for the development of integrated photonic SiC devices and could be extended to a potentially wide range of applications involving other classes of solid-state quantum emitters.
We present a novel approach for the optical activation of the negatively-charged silicon vacancy ( V
Si
−
) center in ion irradiated silicon carbide (SiC) via ns-pulsed laser annealing in the 234–2180 mJ cm
− 2
energy density range. The laser annealing process is investigated under 355 nm and 532 nm wavelengths at pulse energy densities below the melting threshold and validated by means of Raman spectroscopy and photoluminescence mapping. The combined effect of ns pulsed laser annealing and subsequent thermal treatment is also assessed. The results offer a promising resource for the development of integrated photonic SiC devices and could be extended to a potentially wide range of applications involving other classes of solid-state quantum emitters
Towards a Framework for Self-Evolving Products in Additive Manufacturing
No full textIn critical industrial applications, enhancing existing custom-designed products throughout their life cycles is crucial for improving user value while managing the costs of continuous design iterations. Additive manufacturing has demonstrated significant improvements in product performance through optimized designs, but implementing these improvements in scale requires excessive design efforts. In this paper, we address this gap by presenting a digital twin (DT)-based product design framework for additively manufactured parts that makes the design of the physical parts independently to adapt into their industrial application environments. The methodology is based on efficiently utilizing data in the intersection of application-specific DT model and the CAD (Computer-Aided Design). It is proposed that with the smart utilization of Evolutionary Algorithms (EA), most of the manual labor involved in drafting performance-improving revisions of design of part geometry could be eliminated. The proposition is evaluated from the business model point of view highlighting the potential for novel, mutually beneficial supplier-customer relationships in advanced industrial equipment solutions
Simulation-driven Thermal Analyses for Cultural Heritage Conservation
Get PDFHeritage masonry structures, with their complex geometries and variety of materials, pose significant challenges for accurate digital modelling and simulations for restoration and conservation purposes. In addition, traditional cultural heritage (CH) documentation workflows typically rely on geometric 3D acquisitions, often without supporting material diagnostics or structural insights. This work addresses this gap by introducing a data-driven workflow that leverages image-based information and 3D point clouds to extract both geometric and material-related attributes for energy analysis applications. The proposed methodology leverages Deep Learning (DL) and Finite Element Method (FEM) modelling to support energy simulation for cultural heritage assets. The proposed workflow integrates orthoimages and 3D data to segment masonry textures, estimate wall thickness, and generate a semantically enriched mesh tailored for energy analyses. A YOLO-based model identifies stone and mortar regions in high-resolution imagery, while point cloud voxelization and plane fitting are used to compute local thickness values. This information feeds into an adaptive meshing strategy, where mesh resolution is adjusted based on material texture and geometric features. A tunable parameter β enables control over mesh density, allowing for optimization of computational performance in thermal FEM simulations. This approach enables the derivation of meaningful simulation-ready 3D models from limited survey data
FreeInsert: Disentangled text-guided object insertion in 3D Gaussian scene without spatial priors
No full textFrancesco Pellegrini. Medico, storico e umanista veronese - I padri veronesi della medicina RASSEGNA STORICO-ETICA
No full textThis article, published in Verona Medica (History of Medicine section), examines the figure of Francesco Pellegrini (1883–1960), a military physician and historian of medicine whose work exemplifies a close integration of scientific competence, historical inquiry, and ethical responsibility. The paper reconstructs Pellegrini’s medical, military, and scholarly career, with particular attention to his contributions to the history of medicine, his extensive studies on Girolamo Fracastoro, and his reflections on military medicine as a practice of care under conditions of extreme vulnerability. Special emphasis is placed on Pellegrini’s understanding of medicine as a fundamentally humanistic discipline, grounded in clinical observation, proximity to patients, and sensitivity to the moral dimensions of medical practice
A value-sensitive framework for informed consent: considerations from the WHO's Guidance for Human Genome Data Collection, Access, Use and Sharing
Get PDFThe growing complexity of clinical genomic research challenges existing informed consent models, demanding frameworks that reflect the collective, dynamic, and relational nature of genetic data. Drawing on six ethical principles from the recent WHO’s Guidance for Human Genome Data Collection, Access, Use and Sharing — social justice, inclusivity, solidarity, responsible stewardship, transparency, and accountability — this paper proposes Value Sensitive Design as a methodological approach to translate normative aims into practice. It complements models like dynamic consent by integrating stakeholder values from the outset and enhancing ethical robustness, cultural sensitivity, and adaptability across diverse sociotechnical contexts. It also opens pathways for developing digital tools and speculative strategies for inclusive and forward-looking genomic data governance
How to Connect Speech Foundation Models and Large Language Models? What Matters and What Does Not
No full textThe remarkable performance achieved by Large Language Models (LLM) has driven research efforts to leverage them for a wide range of tasks and input modalities. In speech-to-text (S2T) tasks, the emerging solution consists of projecting the output of the encoder of a Speech Foundational Model (SFM) into the LLM embedding space through an adapter module. However, no work has yet investigated how much the downstream-task performance depends on each component (SFM, adapter, LLM) nor whether the best design of the adapter depends on the chosen SFM and LLM. To fill this gap, we evaluate the combination of 5 adapter modules, 2 LLMs (Mistral and Llama), and 2 SFMs (Whisper and SeamlessM4T) on two widespread S2T tasks, namely Automatic Speech Recognition and Speech Translation. Our results demonstrate that the SFM plays a pivotal role in downstream performance, while the adapter choice has moderate impact and depends on the SFM and LLM
Fusion of Airborne and Spaceborne Thermal Imagery for Temperature Monitoring in Urban Areas
No full textLand use and surface properties directly influence and affect events and phenomena, such as flash flooding due to heavy rains, Urban Heat Island (UHI) intensification during heat waves, biodiversity reduction, etc. To deeply study such phenomena within urban areas, remotely sensed data are essential to describe, investigate, and model them and drive actionable insights. Aerial sensors mounted on airplanes allow for the acquisition of high-resolution data in terms of geometry (up to 5 cm with multispectral cameras) and spectral content (hundreds of narrow bands with wavelengths from visible to short-wave and thermal infrared). In urban applications, images are generally acquired during aerial surveys planned on demand by the municipalities. Satellite images, on the other hand, feature a consistent revisit time thanks to their elliptic sun-synchronous orbits. Satellite images with low spatial granularity, often acquired through international public missions (i.e., Copernicus), are publicly accessible. Within the USAGE—Urban Data Space for Green Deal—EU project [https://www.usage-project.eu/], we investigate the integration of multi-modal and multi-resolution data from aerial and satellite platforms in urban areas for environmental analyses. In this paper, the activities specifically realized in thermal analysis are discussed. Two pilot cities are considered, Graz (Austria) and Ferrara (Italy), where aerial multispectral, thermal, hyperspectral, and LiDAR data, as well as thermal satellite images, are available (Beber et al., Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLVIII-1/W3-2023, 9–16, 2023). Firstly, the land surface temperatures (LST) are calculated from thermal aerial images (0.5–1 m spatial resolution), also with the support of aerial hyperspectral images (VNIR and SWIR ranges, 1 m spatial resolution) to retrieve information on the type of surface material, thus on the surface emissivity. The LST values are then compared to those retrieved using the Landsat TIRS sensor, in order to characterize the representativeness of the Landsat pixel over the urban landscape. Moreover, a deep learning algorithm to downscale a Landsat product to airborne resolution is presented. Finally, LST maps are coupled with population density to highlight areas with higher risks. The proposed methodology could be replicated also in other similar cities