University of Modena and Reggio Emilia
Archivio istituzionale della ricerca - Università di Modena e Reggio EmiliaNot a member yet
116400 research outputs found
Sort by
Assessing the Impact of Cybersecurity Attacks on Digital Twin Metrics: An Experimental Study
No full textDielectric Absorption Caused by Traps in MIM/MOS Capacitors: A New Model Validated Through TCAD
No full textWe propose here a flexible physics-based model to calculate the capacitance and conductance in MIM and MOS capacitors, accounting for traps with different energy and space distributions in the dielectric. The proposed model is validated against calibrated TCAD simulations. Furthermore, the frequency and temperature dependencies of the loss tangent obtained from the TCAD simulations have been analyzed and compared with experimental trends reported in the literature, highlighting their relation with the trap location in the oxide and the electron trapping process. These findings offer insight into the trap-induced effects in MIM capacitors, helping to determine the physical origin of the dielectric absorption (DA) phenomenon affecting relevant CMOS circuits such as analog-to-digital converters (ADCs)
Forecasting Irregularly Sampled Time Series with Transformer Encoders
No full textTime series forecasting is a fundamental task in various domains, including environmental monitoring, finance, and healthcare. State-of-the-art forecasting models typically assume that time series are uniformly sampled. However, in real-world scenarios, data is often collected at irregular intervals and with missing values, due to sensor failures or network issues. This makes traditional forecasting approaches unsuitable. In this paper, we introduce ISTF (Irregular Sequence Transformer Forecasting), a novel transformer-based architecture designed for forecasting irregularly sampled multivariate time series (MTS). ISTF leverages exogenous variables as contextual information to enhance the prediction of a single target variable. The architecture first regularizes the MTS on a fixed temporal scale, keeping track of missing values. Then, a dedicated embedding strategy, based on a local and global attention mechanism, aims at capturing dependencies between timestamps, sources and missing values. We evaluate ISTF on two real-world datasets, FrenchPiezo and USHCN. The experimental results demonstrate that ISTF outperforms competing approaches in forecasting accuracy while remaining computationally efficient
Ultra long-term EEG monitoring for developmental and epileptic encephalopathies: protocol for a prospective study using subscalp EEG
No full textFG-TRACER: Tracing Information Flow in Multimodal Large Language Models in Free-Form Generation
No full textMultimodal Large Language Models (MLLMs) have achieved impressive performance across a variety of vision–language tasks. However, their internal working mechanisms remain largely underexplored. In his work, we introduce FG-TRACER, a framework designed to analyze the information flow between visual and textual modalities in MLLMs in free-form generation. Notably, our numerically stabilized computational method enables the first systematic analysis of multimodal information flow in underexplored domains such as image captioning and chain-of-thought (CoT) reasoning. We apply FG-TRACER to two state-of-the-art MLLMs—LLaMA 3.2-Vision and LLaVA 1.5—across three vision–language benchmarks—TextVQA, COCO 2014, and ChartQA—and we conduct a word-level analysis of multimodal integration. Our findings uncover distinct patterns of multimodal fusion across models and tasks, demonstrating that fusion dynamics are both model- and task-dependent. Overall, FG-TRACER offers a robust methodology for probing the internal mechanisms of MLLMs in free-form settings, providing new insights into their multimodal reasoning strategies. Our source code is publicly available at https://anonymous.4open.science/r/FG-TRACER-CB5A/
The integration of gene mutations and copy number variations refines the prognosis of mantle cell lymphoma: long-term results of the Fondazione Italiana Linfomi MCL0208 clinical trial
No full textSketch2Stitch: GANs for Abstract Sketch-Based Dress Synthesis
No full textIn the realm of creative expression, not everyone possesses the gift of effortlessly translating their imaginative visions into flawless sketches. More often than not, the outcome resembles an abstract, perhaps even slightly distorted representation. The art of producing impeccable sketches is not only challenging but also a time-consuming process. Our work is the first of this kind in transforming abstract, sometimes deformed garment sketches into photorealistic catalog images, to empower the everyday individual to become their own fashion designer. We create Sketch2Stitch, a dataset featuring over 65,000 abstract sketch images generated from garments of DressCode and VITONHD, two benchmark datasets in the virtual try-on task. Sketch2Stitch is the first dataset in the literature to provide abstract sketches in the fashion domain. We propose a StyleGAN-based generative framework that bridges freehand sketching with photorealistic garment synthesis. We demonstrate that our framework allows users to sketch rough outlines and optionally provide color hints, producing realistic designs in seconds. Experimental results demonstrate, both quantitatively and qualitatively, that the proposed framework achieves superior performance against various baselines and existing methods on both subsets of our dataset. Our work highlights a pathway toward AI-assisted fashion design tools, democratizing garment ideation for students, independent designers, and casual creators
Artificial intelligence in healthcare: Proposal for a new medico-legal methodology in medical liability
No full textThe rapid integration of Artificial Intelligence (AI) into healthcare promises significant benefits but also raises unprecedented ethical, clinical, and legal challenges. Current medico-legal frameworks, primarily designed for human decision-making, are often inadequate to address liability issues arising from algorithmic errors or opaque "black box" models. This paper introduces a novel medico-legal methodology that combines proactive and reactive approaches to risk assessment, originally developed within European forensic medicine, and adapts it to the context of AI in healthcare. By systematically analyzing data collection, dataset validation, error identification, and causal reconstruction, the proposed framework provides a structured path for evaluating medical liability when AI systems are involved. This dual approach not only supports clinicians, developers, and policymakers in preventing harm, but also establishes a robust forensic tool for liability assessment. The methodology offers a step toward internationally applicable standards for addressing the medico-legal implications of AI in medicine
Assessment of dual-oxide options for LDMOS transistors in FinFET technology
No full textWe present a comparison of three LDMOS transistor designs in ≃16 nm FinFET technology with
different gate stack configurations: thick ITL oxide, thin ITL oxide, and a combination of both
(dual ITL oxide thickness). We analyze by simulation how the gate oxide stack influences the main
performance and time-zero degradation rate indicators.
The simulations suggest that, for the same doping profile and gate length (LG), the dual-oxide
configuration has transition frequency (fT) and on-resistance (RDS,on) within ≈16% and 6% of those
of thick and thin-oxide devices, respectively, while the maximum substrate and gate currents are
≈35% and 43% smaller than for a fully thin-oxide device, respectively. Consequently, the dual-oxide
configuration enables LG scaling and improvements in fT and RDS,on while keeping degradation
monitors under control