Archivio della ricerca - Fondazione Bruno Kessler
Not a member yet
21227 research outputs found
Sort by
TVEG: Model Selection of the Time-Varying Exponential Family Distributions Graphical Models
No full textThe undirected graphical model, a popular class of statistical model, offers a way to describe and explain the relationships among a set of variables. However, it remains a challenge to choose a certain graphical model to explain the relationships of variables adequately, especially when the relationships of variables are rewiring over time. This paper proposes the Time-Varying Exponential Family Distributions Graphical (TVEG) models, with time-varying structures and exponential family node-wise conditional distributions. TVEG models extend the scope of available graph models and can be applied to time-varying and exponential family distribution observation data in reality. We propose the Temporally Smoothed L1-regularized exponential family graphical estimator (TSLEG), an estimator to infer the structure of TVEG from observations. We derive sufficient conditions for the TSLEG to recover the block partition and sparse pattern with high probability. We derive a message-passing optimization method to solve the TSLEG for time-varying Ising, Gaussian, exponential, and Poisson graphs based on the ADMM. The synthetic network simulations corroborate the theoretical analysis. Analysing of real data of stocks and the US Senate by the time-varying exponential model and Poisson model indicates the effectiveness and practicality of TVEG models
ModaFact: Multi-paradigm Evaluation for Joint Event Modality and Factuality Detection
Get PDFFactuality and modality are two crucial aspects concerning events, since they convey the speaker`s commitment to a situation in discourse as well as how this event is supposed to occur in terms of norms, wishes, necessity, duty and so on. Capturing them both is necessary to truly understand an utterance meaning and the speaker`s perspective with respect to a mentioned event. Yet, NLP studies have mostly dealt with these two aspects separately, mainly devoting past efforts to the development of English datasets. In this work, we propose ModaFact, a novel resource with joint factuality and modality information for event-denoting expressions in Italian. We propose a novel annotation scheme, which however is consistent with existing ones, and compare different classification systems trained on ModaFact, as a preliminary step to the use of factuality and modality information in downstream tasks. The dataset and the best-performing model are publicly released and available under an open license
Virtual Reality (VR) for Neurodegenerative Disorders: Key Findings and Future Directions
No full textIntroduction: Virtual Reality (VR) is emerging as a non-pharmacological tool in healthcare with significant potential for addressing cognitive and behavioral symptoms of aging and neurodegenerative diseases. Its immersive and customizable nature offers novel therapeutic opportunities, particularly for dementia-related conditions. This presentation summarizes findings from four studies exploring VR applications and their role in enhancing patient-centered care for neurodegenerative disorders. The first study evaluated personalized VR scenarios for older adults with cognitive impairments, demonstrating high engagement, relaxation, and positive reminiscence effects. Minimal discomfort was noted, though improvements in wearability and usability are needed. Building on this, the second study examined how tailored multi-sensory VR environments help manage behavioral and psychological symptoms of dementia by reducing agitation and promoting emotional well-being. In the third study, we explored VR-based Reminiscence Therapy for individuals with Parkinson’s-related mild cognitive impairment, integrating AI-generated visuals and adaptive storytelling to foster emotional engagement, reduce depressive symptoms, and improve quality of life. The fourth study assessed the integration of VR and digital tools within clinical settings, identifying usability challenges, adoption barriers, and ethical considerations.
Methods: Participants included 23 dementia patients (study 1), 20 inpatients with cognitive impairment (study 2), and 20 outpatients with Parkinson’s-related mild cognitive impairment (study 3). The fourth study involved 10 healthcare staff members as stakeholders. Data collection relied on self-reports, observational tools, and focus groups with patients and healthcare professionals.
Results: The initial study showed high feasibility and acceptability, with VR enabling the creation of personalized environments beyond real-world constraints. Preliminary findings from ongoing studies highlight VR’s capacity to enhance emotional well-being, reduce agitation, and provide meaningful engagement.
Conclusions and Future Directions: VR demonstrates promising potential to improve cognitive, emotional, and behavioral outcomes for neurodegenerative conditions. Future research will focus on refining interventions, improving accessibility, expanding AI-driven personalization, and conducting large-scale trials to establish evidence-based practices in dementia care
Green hydrogen and renewable energy comunities: synergies for a sustainable future
No full textThe interaction and coexistence of hydrogen production and Renewable Energy Communities (RECs) form a critical nexus and a promising pathway toward sustainable energy systems [1]. As the world faces the twin challenges of climate change and energy security, hydrogen has emerged as a versatile energy carrier that facilitates the integration of renewable energy sources (RES). Integrating green hydrogen technologies into RECs represents a unique opportunity to develop sustainable and resilient energy systems, accelerating the global transition to a low-carbon economy. By converting surplus renewable energy into hydrogen, the stored hydrogen can be blended with gas/biogas to fuel industrial processes within companies. Beyond their high electrical efficiency, stationary fuel cells also generate heat, which can be exploited for use by local industries and small businesses. However, under the current REC regulations in Italy, integrating hydrogen into RECs remains challenging. While the use of hydrogen to produce electricity is not recognized as incentivable, there is no explicit restriction on the use of green hydrogen and the fact that an electrolyser can be a consumer within a REC, using excess PV-generated electricity to produce hydrogen. Exploring the use of a rSOC (reversible Solid Oxide Cell) in FC mode is possible to use biogas in case of hydrogen unavailability to produce incentivable electricity for the REC. The resulting green hydrogen from EL (Electrolyser) mode of a rSOC could serve as a long-term energy storage solution, addressing the intermittency challenges of renewable energy sources. This study explores a possible scenario to demonstrate the feasibility of establishing a sustainable and cost-effective synergy between hydrogen and RECs, focusing on key challenges and critical legislative barriers arising from their interaction. As technological advancements continue to drive down costs and improve efficiency, the growing adoption of green hydrogen and RECs has the potential to significantly impact the energy landscape in the future
Fine-grained Fallacy Detection with Human Label Variation
Get PDFWe introduce FAINA, the first dataset for fallacy detection that embraces multiple plausible answers and natural disagreement. FAINA includes over 11K span-level annotations with overlaps across 20 fallacy types on social media posts in Italian about migration, climate change, and public health given by two expert annotators. Through an extensive annotation study that allowed discussion over multiple rounds, we minimize annotation errors whilst keeping signals of human label variation. Moreover, we devise a framework that goes beyond “single ground truth” evaluation and simultaneously accounts for multiple (equally reliable) test sets and the peculiarities of the task, i.e., partial span matches, overlaps, and the varying severity of labeling errors. Our experiments across four fallacy detection setups show that multi-task and multi-label transformer-based approaches are strong baselines across all settings. We release our data, code, and annotation guidelines to foster research on fallacy detection and human label variation more broadly
Learning-based 3D reconstruction methods for non-collaborative surfaces-A metrological evaluation
Get PDFNon-collaborative (i.e., reflective, transparent, metallic, etc.) surfaces are common in industrial production processes, where 3D reconstruction methods are applied for quantitative quality control inspections. Although the use or combination of photogrammetry and photometric stereo performs well for well-textured or partially textured objects, it usually produces unsatisfactory 3D reconstruction results on non-collaborative surfaces.
To improve 3D inspection performances, this paper investigates emerging learning-based surface reconstruction methods, such as Neural Radiance Fields (NeRF), Multi-View Stereo (MVS), Monocular Depth Estimation (MDE), Gaussian Splatting (GS) and image-to-3D generative AI as potential alternatives for industrial inspections. A comprehensive evaluation dataset with several common industrial objects was used to assess methods and gain deeper insights into the applicability of the examined approaches for inspections in
industrial scenarios. In the experimental evaluation, geometric comparisons were carried out between the reference data and learning-based reconstructions. The results indicate that no method can outperform all the others across all evaluations
Revisiting the properties of superfluid and normal liquid 4He using ab initio potentials
No full textWe investigate the properties of liquid
He in both the normal and superfluid phases using path-integral Monte Carlo simulations and recently developed ab initio potentials that incorporate pair, three-body, and four-body interactions. By focusing on the energy per particle as a representative observable, we use a perturbative approach to quantify the individual contributions of the many-body potentials and systematically propagate their associated uncertainties. Our findings indicate that the three-body and four-body potentials contribute to the total energy by approximately 4% and 0.5%, respectively. However, the primary limitation in achieving highly accurate first principles calculations arises from the uncertainty in the four-body potential, which currently dominates the propagated uncertainty. In addition to the energy per particle, we analyze other key observables, including the superfluid fraction, condensed fraction, and pair distribution function, all of which demonstrate excellent agreement with experimental measurements
Sensore di gas a stato solido e corrispondente procedimento di fabbricazione
No full textObject of the invention
The object of the present invention is to provide an improved solid-state gas sensor that allows achieving higher selectivity than known gas sensors,
solving one or more of the technical problems mentioned above
Ionized Jet Deposition of MoS2 on Gas Diffusion Layer Electrodes for Next Generation Alkaline Electrolyzers
No full textThis study focuses on optimizing MoS2 catalysts for the hydrogen evolution reaction (HER) in anion exchange membrane (AEM) electrolyzers. A scalable Ionized Jet Deposition (IJD) technique is employed to deposit MoS2 onto various carbon supports, exploring the relationship between substrate properties and catalytic performance. The results demonstrate that substrate choice plays a pivotal role in enhancing HER activity and durability. MoS2 deposited on Freudenberg carbon support exhibited the best catalytic activity, achieving a current density of 10 mA μg−1Mo at −0.48 V versus RHE in an alkaline environment, even with a low catalyst loading (12–49 μg cm−2). Conversely, sulfur-doped carbon supports showed lower HER activity but superior stability, with a minimal voltage degradation of just 0.025 V after 6 h of testing at 10 mA cm−2. To further understand these results, bubble evolution studies, and contact angle measurements are conducted. Stable electrodes demonstrated small contact angles and enhanced bubble release from the surface, indicating the importance of hydrophilicity in improving performance and durability. This work highlights the synergy between scalable synthesis techniques and substrate optimization, offering a promising path for advancing cost-efficient, durable electrocatalysts in large-scale AEM electrolyzers for green hydrogen production
A Proactive Decoy Selection Scheme for Cyber Deception using MITRE ATT&CK
No full textCyber deception allows compensating the late response of defenders countermeasures to the ever evolving tactics, techniques, and procedures (TTPs) of attackers. This proactive defense strategy employs decoys resembling legitimate system components to lure stealthy attackers within the defender environment, slowing and/or denying the accomplishment of their goals. In this regard, the selection of decoys that can expose the techniques used by malicious users plays a central role to incentivize their engagement. However, this is a difficult task to achieve in practice, since it requires an accurate and realistic modeling of the attacker capabilities and his possible targets. In this work, we tackle this challenge and we design a decoy selection scheme that is supported by an adversarial modeling based on empirical observation of real-world attackers. We take advantage of a domain-specific threat modeling language using MITRE ATT&CK© framework as source of attacker TTPs targeting enterprise systems. In detail, we extract the information about the execution preconditions of each technique as well as its possible effects on the environment to generate attack graphs modeling the adversary capabilities. Based on this, we formulate a graph partition problem that minimizes the number of decoys detecting a corresponding number of techniques employed in various attack paths directed to specific targets. We compare our optimization-based decoy selection approach against several benchmark schemes that ignore the preconditions between the various attack steps. Results reveal that the proposed scheme provides the highest interception rate of attack paths using the lowest amount of decoys