Archivio della ricerca - Fondazione Bruno Kessler
Not a member yet
21227 research outputs found
Sort by
Beam test preliminary results of the ADA_5D LGADs Detectors and Front-End electronics
No full textCharge identification is a challenging task in space-based cosmic-ray (CR) experiments. This is due to the wide dynamic range required to identify CR elements, including heavy nuclei. Additionally, back-scattered radiation from the calorimeter degrades the charge resolution when it hits the same detector element traversed by the cosmic ray, hindering a correct identification of the cosmic nucleus.
This issue will be particularly relevant for future experiments aimed at exploring the highest energy part of the CR spectrum, as the amount of back-scattering increases with particle energy.
To address these challenges, the ADA_5D project, funded by the Italian National Institute for Nuclear Physics (INFN), is developing an innovative detector, based on arrays of Low Gain Avalanche Diode (LGAD) pixels. It is designed to be capable of simultaneously measuring position, charge (with a very wide dynamic range, up to Z ∼ 40), and timing, with sub-nanosecond resolution. It uses a scalable technology suitable for covering wide areas (∼m2) with low power consumption, making it ideal for space experiments.
This paper reports the first beam test results of the LGADs and front-end electronics being developed for the ADA_5D project. The test was carried out at the CERN-SPS North Area facility, during the 2024 winter campaign with Pb ions.
The beam test setup included a controlled environment where the prototypes were exposed to high-energy Pb fragment beams.
The performance of the detector and of the front-end electronics was evaluated in terms of signal response and timing resolution. Preliminary results show promising performance, with the ADA_5D chip demonstrating stable operation under high-rate conditions
Exploring Paraphrasing Strategies for CEFR A1-Level Constraints in {LLM}s
Get PDFLarge language models are increasingly used for teaching and self-learning foreign languages. However, their capability to meet specific linguistic constraints is still underexplored. This study compares the effectiveness of prompt engineering in guiding ChatGPT (4o and 4o-mini), and Llama 3 to rephrase general-domain texts to meet CEFR A1-level constraints in English and Italian, making them suitable for beginner learners. It compares 4 prompt engineering approaches, built upon iterative paraphrasing method that gradually refines original texts for CEFR compliance. The approaches compared include paraphrasing with or without Chain-of-Thought, as well as grammar and vocabulary simplification performed either simultaneously or as separate steps. The findings suggest that for English the best approach is combining COT with separate grammar and vocabulary simplification, while for Italian one-step strategies have better effect on grammar, and two-step strategies work better for covering the vocabulary. The paraphrasing approach can approve compliance, although at this point it is not cost-effective. We release a dataset of pairs original sentence-beginner level paraphrase (both in Italian and in English) on which further work could be based
A Review on Trustworthiness of Digital Assistants for Personal Healthcare
No full textArtificial Intelligence (AI) is widely used within the healthcare domain. One of the branches of digital health concerns the design and development of digital assistant solutions. AI-enabled digital assistants highlight the need to be trustworthy given their intrusiveness within people’s lives. Such solutions aim to provide intelligent tools to ease the management of care pathways or to enhance the capabilities of healthcare organizations in deploying health prevention campaigns by monitoring the lifestyles of healthy people. In this work, we intend to analyze the recent literature concerning integrating AI techniques within digital assistants. We focus on the contributions published during the last 10 years and perform a careful analysis of whether and how trustworthy pillars have been addressed. We also discuss the risks of designing digital assistants without considering trustworthy pillars and present some recommendations to mitigate them
Automatic detection of speech sound disorders in German-speaking children: augmenting the data with typically developed speech
Get PDFSpeech Sound Disorders (SSD) are common among children, affecting their academic, social, and emotional development. Traditional diagnostic methods are based on speech-language pathologists, making them resource-intensive. Due to the global shortage of experts and increasing demand, exploring deeplearning tools is crucial. Adapting a multi-task framework to f ine-tune a pre-trained multilingual Wav2Vec model, this study tackles Automatic Speech Recognition and SSD classification for German children using a custom dataset. We show that incorporating public out-of-domain datasets improves robustness and generalizability. Interestingly, combining pathological and typical speech data with mispronunciations benefits the performance in terms of speech recognition and SSD detection. Finally, we investigate a two-step training of the model that further improves the overall performance
Beyond Human-Centric Play: A Review of Commercial Video Games to Inform More-than-Human Serious Game Design
No full textA TinyMLOps-Based Edge AI Approach for Early Detection of Emerging Plant Diseases
No full textThis paper presents a computer vision-based system for the early detection of plant leaf diseases, developed using TinyMLOps principles to enable deployment on resourceconstrained embedded devices. The solution leverages the MobileNet architecture with a custom classification head that is properly trained using the PlantVillage dataset to distinguish between healthy, unhealthy, and no leaves images. Given the class imbalance of the relabeled dataset, we augmented the underrepresented classes (No leaves) using basic image transformations to improve model generalization. To ensure both high accuracy and computational feasibility, considering that the target device has limited resources, we apply a multi-objective hyperparameter optimization strategy using the Optuna framework, incorporating the NSGA-II algorithm to identify optimal trade-offs between model accuracy (measured by F1-score) and computational constraints (number of model parameters). The optimization process identifies a set of Pareto-optimal models, with the most promising configurations residing in the “knee region”, offering strong classification performance with significantly reduced parameter counts. From the Pareto front, we selected a few models that were evaluated on a Raspberry Pi Zero 2W to assess their realworld performance in terms of inference latency, memory usage, and overall suitability for deployment in resource-constrained environments. This poses the basis for building a low-cost solution with commodity devices for early disease detection in agriculture, including rural areas where computational resources and connectivity are limited. Finally, the dataset we used for this work has been made public as an asset of the AgrifoodTEF Data Space
First-AID: the first Annotation Interface for grounded Dialogues
Get PDFThe swift advancement of Large Language Models (LLMs) has led to their widespread use across various tasks and domains, demonstrating remarkable generalization capabilities. However, achieving optimal performance in specialized tasks often requires fine-tuning LLMs with task-specific resources. The creation of high-quality, human-annotated datasets for this purpose is challenging due to financial constraints and the limited availability of human experts. To address these limitations, we propose First-AID, a novel human-in-theloop (HITL) data collection framework for the knowledge-driven generation of synthetic dialogues using LLM prompting. In particular, our framework implements different strategies of data collection that require different user intervention during dialogue generation to reduce post-editing efforts and enhance the quality of generated dialogues. We also evaluated First-AID on misinformation and hate countering dialogues collection, demonstrating (1) its potential for efficient and high-quality data generation and (2) its adaptability to different practical constraints thanks to the three data
collection strategies
Bypassing Duty-Cycle Limitations for RL-Enhanced LoRaWAN Communications
No full textThe rise of IoT in the past few years has led to the massive deployment of connected devices making IoT networks denser. To optimize transmission arising from congestion in dense networks, adaptive data-rate algorithms have been implemented such as the one used in the LoRaWAN protocol. Utilization of algorithms based on reinforcement learning, especially multi-armed bandit, have been investigated, but the duty-cycle limitation decreases the performance of these algorithms in dense networks up to 15 %. This paper aims at giving a solution to resolve the issue caused by duty-cycle limitation, using the LoRa technology as a study case. An effort was done on energy consumption and the reward is modified in order to save energy according to the quality of service. Performances are evaluated thanks to our new simulator J-LoRaNeS based on the Julia language
ZWISCHEN WÄLDERN UND GEWÄSSERN. Eine Neubetrachtung der Welschen Konfinen in der frühen Neuzeit
No full textAt the beginning of the 16th century, the Alpine region of Trentino-Tyrol was one of the most important areas of the Upper Austrian Archduchy for the extraction of natural resources due to the abundance of forests and mines and, in particular, waterways, which facilitated the rapid transport of these goods to the sales markets. This article analyzes the environmental impact of timber exploitation in the Cismon-Brenta river basin in the Welschen Konfinen district, focusing on the analysis of the infrastructure for timber rafting, social relations, and political governance associated with managing the complexity of the „timber frontier“ between the Holy Roman Empire and the Republic of Venice, one of the most anthropized areas in early modern Europ