107187 research outputs found
Sort by
Integrated mechanical and electromagnetic geophysical imaging of a complex aquifer system: the coastal plain in Muravera, south-east Sardinia (Italy)
No full textWater management is crucial to lead sustainable development of our societies, and to be effective needs for a detailed characterization of the water resources availability, through the knowledge of their reservoir’s main properties in terms of geometries and physical properties. In this way it is mandatory to plan responsible exploitation and use protocols, increasing the system resilience. Usable groundwater scarcity is in fact a major issue to be overcome with smart territorial management. In view of this, the knowledge derived from direct and indirect diagnostic investigations is a key factor to balance the societal pressure over such delicate targets. In the early Twenty-thousands, a geophysical survey using High-Resolution (HR) reflection seismic and time domain electromagnetics (TDEM) was carried out in the south-eastern part of the island of Sardinia, Italy. The main objective of the survey has been to study a very complex coastal aquifer situated in an area where a – still present – huge saltwater intrusion phenomenon had already occurred mainly due to anthropic overexploitation linked to tourism and agriculture. Within this paper, the TDEM results, constrained by seismics, are presented and discussed
Phase angle and vector analysis in the evaluation of body composition in sarcopenic obesity: a systematic review
No full textBackground Sarcopenic obesity (SO) is a condition characterized by low muscle mass and strength and high adiposity. Bioelectrical impedance analysis (BIA) derived phase angle (PhA) and bioelectrical impedance vector analysis (BIVA) are simple and inexpensive tools for the evaluation of body composition, with an emerging consensus in health research and application. Aim The aim of this systematic review was to analyze research on sarcopenic obesity using PhA or BIVA. Methods A bibliographic search was performed on 13 January 2025, using three databases: PubMed, Scopus and Web of Science. The search terms were: ("phase angle" OR BIVA) and (sarcopenic OR sarcopenia OR obesity). Studies addressing only obesity or sarcopenia were excluded. The quality of the studies was evaluated using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies, from the National Institute of Health. No meta-analysis was conducted. Results Nine studies were selected, mostly published in 2022 (89%) or later and focused on clinical applications (55.6%). The reviewed studies showed substantial methodological variability. Diagnostic criteria included the ESPEN–EASO algorithm as well as protocols based on different definitions of sarcopenia and obesity. Indices and cut-offs used to define body composition varied accordingly. Variability was also observed in population samples and in bioimpedance devices. All selected studies used PhA and two of them used BIVA. Although quantitative results are variable, with PhA values ranging from 3.9° to 7.1°, and mostly below 5.6°. The qualitative pattern of bioelectrical characteristics associated with body composition in SO is broadly consistent across studies: PhA is tendentially lower than in healthy subjects and patients with obesity and similar to those with sarcopenia; the specific vector is longer. Conclusions Research is still quite heterogeneous in terms of methods and diagnostic procedures, which limits the comparability of the results. However, the observed tendencies confirm the suitability of PhA for recognizing the reduced muscle mass associated with sarcopenia, while specific BIVA also appears capable of detecting excess fat mass related to obesity. Further research is needed to standardize procedures for characterizing sarcopenic obesity and monitoring its progression
Reasoned or Rapid code? Unveiling the strengths and limits of DeepSeek for Solidity development
No full textContext:
As blockchain systems grow in complexity, secure and efficient smart contract development remains a crucial challenge. Large Language Models (LLMs) like DeepSeek promise significant enhancements in developer productivity through automated code generation, debugging, and testing. This study focuses on Solidity, the dominant language for Ethereum smart contracts, where correctness, gas efficiency, and security are critical to real-world adoption.
Objective:
This study evaluates the capabilities of DeepSeek’s V3 and R1 models, a non-reasoning Mixture-of-Experts architecture and a reasoning-based model trained via reinforcement learning, respectively, in automating Solidity contract generation and testing, as well as identifying and fixing common vulnerabilities.
Methods:
We designed a controlled experimental framework to evaluate both models by generating and analysing a diverse set of smart contracts, including standardised tokens (ERC20, ERC721, ERC1155) and real-world application scenarios (Supply Chain, Token Exchange, Auction). The evaluation is grounded on a multidimensional metric suite covering quality, technical robustness and process characteristics. Vulnerability detection and patching capabilities are tested using predefined vulnerable contracts and guided patch prompts. The analysis spans six levels of prompt complexity and compares the impact of reasoning-based and non-reasoning-based generation strategies.
Results:
Findings reveal that R1 delivers more accurate and optimised outputs under high complexity, while V3 performs more consistently in simpler tasks with simpler code structures. However, both models exhibit persistent hallucinations, limitations in vulnerability coverage, and inconsistencies due to prompt formulation. The correlation between re-evaluation patterns and output quality suggests that reasoning helps in complex scenarios, although excessive revisions may lead to over-engineered or unstable solutions.
Conclusions:
Neither model is robust enough to autonomously generate issue-free smart contracts in complex or security-critical scenarios, underscoring the need for human oversight. These findings highlight best practices for integrating LLMs into blockchain development workflows and emphasise the importance of aligning model selection with task complexity and security requirements
Race against time: investigating the factors that influence web race condition exploits
No full textRace conditions (RC) pose a critical security threat to web applications by exploiting the non-deterministic behavior of multithreaded request handling. This can lead to unpredictable outcomes such as data corruption, Time of Check to Time of Use (TOCTOU) vulnerabilities, and deadlocks. While previous research has identified poor design practices that contribute to RC vulnerabilities, no existing studies have explored the factors that influence the severity or impact of race conditions. This paper introduces a comprehensive methodology for testing and quantifying how different variables affect the exploitability of race conditions in vulnerable web servers, providing a framework for future research to investigate this issue more thoroughly. In addition, we present an experimental evaluation of our methodology under various conditions. Specifically, we examine six RC exploitation tools using four different attack techniques across both HTTP/1.1 and HTTP/2 protocols. To provide a complete overview of race conditions across all HTTP versions, we also introduce the first race condition attack tool for HTTP/3, named QUICker. Furthermore, we assess how the choice of database management systems and programming languages used in web application deployment can affect susceptibility to race condition attacks. This study offers key insights into how these factors influence the exploitability of RC vulnerabilities
A Transformer-Based Framework for OCT Cyst Segmentation
No full textOptical Coherence Tomography is a widely used imaging modality for diagnosing and monitoring retinal diseases. Automated segmentation of cystoid spaces in retinal scans is crucial for early detection and personalized treatment planning. In this work, we propose a novel transformer-based framework that leverages SegFormer for accurate segmentation of cystoid regions in OCT images. The study evaluates the impact of model size and different augmentation strategies, such as geometric and color-based transformations, on the segmentation performance. Our experiments, conducted using the OCT-Cyst Segmentation Challenge dataset, achieve high scoring results, with the best configuration yielding a Dice score of 84.57. Additionally, we provide qualitative analyses to highlight the strengths and limitations of the framework in real-world scenarios. The results demonstrate the robustness and clinical relevance of transformer-based architectures in OCT cyst segmentation. The code associated with this manuscript is available at: https://github.com/unica-visual-intelligence-lab/OCT-Cyst-Segmentation
Event-GPT: Sequence-Aware Video Event Classification via LoRA-Tuned GPT
No full textFire and smoke detection in outdoor environments poses critical challenges for public safety and environmental protection. Traditional single-frame detection methods often fail to capture the temporal dynamics essential for early fire event recognition. This paper introduces Event-GPT, a novel hybrid architecture that combines VideoMAEv2 encoder with a LoRA-tuned GPT-2 decoder for temporally-aware fire and smoke event classification. Our approach processes video sequences through an iterative fusion strategy that accumulates spatiotemporal embeddings across time, enabling robust reasoning over long-term temporal dependencies. Unlike conventional frame-based classifiers, Event-GPT maintains an evolving internal state that captures the progressive nature of fire events. Extensive experiments on the ONFIRE 2025 dataset demonstrate that our method achieves 84.61% precision and 91.66% recall with an average notification delay of 7.14 s when trained at 1 FPS and tested at 4 FPS. The architecture’s memory-efficient design, enabled by LoRA adaptation, ensures real-time performance while maintaining high detection accuracy. The code associated with this manuscript is available at: https://github.com/unica-visual-intelligence-lab/Event-GP
Zum Nutzen altsprachlicher Grundkenntnisse für die fachkommunikative Kompetenz. Ein mehrsprachiger Ansatz für den DaF-Unterricht im medizinischen Bereich
No full textIn today’s society, the motivation to learn German is closely linked to educational and professional goals. This leads to a significant increase in the demand for specialised German as a Foreign Language courses. Focusing on the medical profession, this paper examines the growing demand for foreign medical professionals with German language skills and highlights the enduring role of classical languages, namely Greek and Latin, in medical terminology. It also explores the role of inferential strategies and of cross-linguistic competence in enhancing morphological awareness and improving overall communicative competence, even for beginners, thus contributing to the broader discussion about teaching German for specific purposes
Memoria
No full textLa memoria è un concetto fondamentale che attraversa diverse discipline, quali filosofia, psicologia e neuroscienze, e che ha suscitato interesse e riflessione per secoli. Questo contributo si propone di rispondere alla domanda “Che cos’è la memoria?” attraverso un percorso storico e teorico, partendo dalle tradizioni filosofiche antiche e moderne, passando per le principali teorie psicologiche del Novecento, fino ad arrivare ai modelli computazionali sviluppati nella seconda metà del secolo scorso. Infine, si
presenta una panoramica sui dibattiti contemporanei, che coinvolgono prospettive innovative come la simulazione mentale, il viaggio mentale nel tempo e l’approccio embodied. Ogni fase contribuisce a delineare la complessità della memoria e la sua importanza nello studio della cognizione
umana
Toward Realistic AI-Generated Student Questions to Support Instructor Training
No full textInstructor effectiveness is fundamental to student learning, with the ability to manage student inquiries serving as a critical component of effective teaching. Student questions represent a valuable training resource for instructors to strengthen their teaching strategies, yet interactions with students are often constrained by several factors. In this paper, we investigate how instructors perceive machine- and student-generated questions, considering the potential for the former to complement the latter in a cost-effective manner. Our study involved 121 undergraduate students and an equivalent number of simulated students modeled using a state-of-the-art large language model, generating over 360 questions in total based on video lectures given by seven university instructors. We assessed whether instructors could distinguish between human- and machine-generated questions and how they evaluated their relevance, clarity, answerability, challenge level, and cognitive depth. Results show that instructors struggle to differentiate between the two sets of questions, with accuracy close to random chance. Instructors tended to (i) rate machine-generated questions slightly higher in relevance, clarity, answerability, and challenge—though only relevance and answerability showed significant differences—and (ii) associate them marginally more often with higher-order cognitive skills. This confirms the potential of machine-generated questions as tools for instructor training. Repository: https://github.com/tail-unica/realistic-ai-generated-questions
Integrated Signal Demodulation, Data Conversion, and Packaging Strategies for a Wireless, Fully Implantable, Bidirectional Neural Interface
No full textOver the past decades, an important area within bioengineering has focused on restoring lost physiological function in humans. Among the various application domains, one of the most relevant is the development of upper-limb prostheses to replace missing limbs following amputation. Neural prosthetics represents a promising research direction, with the potential to substantially improve the quality of life of amputees. Unlike commercial prostheses, which are typically controlled through EMG signals and therefore provide limited and unnatural control with no sensory feedback, neural prostheses aim to overcome these limitations by using the same implanted electrodes both for motor control and for restoring a sense of touch.
This thesis presents contributions to the development of a fully implantable, bidirectional, wireless neural interface intended for neuroprosthetic and bioelectronic applications. The system, conceived within a broader collaborative project, consists of an inductively powered implantable hub and an external unit fabricated in 180nm CMOS technology, enabling wireless power transfer and both low- and high-bitrate data communication links. The implantable platform distributes power and data to multiple distributed recording and stimulation front-ends. Within this framework, the thesis focuses on the design and implementation of mixed-signal and packaging blocks required for the implantable platform. A compact demodulation and decoding scheme for the wired interface between the implantable hub and the distributed neural front-ends is presented, enabling simultaneous power and data trans- mission over a four-wire connection through amplitude-modulated clock encoding. The thesis further addresses low-power neural signal acquisition by developing an analog-to-digital conversion scheme that initially follows the Wilkinson architecture, in which the input sample is first converted into a time interval by measuring the discharge time of a capacitor, and this interval is subsequently digitized by a time- to-digital converter (TDC). The TDC is implemented using a two-phase conversion scheme with temporal amplification via current-starved ring oscillators, enabling a significant reduction in clock frequency and energy consumption while maintaining the required resolution for neural recording. Furthermore, the thesis contributes to the early-stage development of a high-density hermetic feedthrough fabrication process using unsintered HTCC and screen-printed platinum, laying the technological groundwork for miniaturized, biocompatible interconnects in implantable packaging