186698 research outputs found
Sort by
Nouvelles recherches et données sur les sites liguriens du Riparo Bombrini (Balzi Rossi, Imperia) et de l’Arma Veirana (Erli, Savona)
No full textProspettive contrastive e acquisizionali sul disapprendimento fra italiano e spagnolo
No full textQuesto studio analizza il fenomeno del disapprendimento nell’acquisizione dell’italiano L2 da parte di parlanti ispanofoni, con particolare attenzione al ruolo dell’esposizione linguistica. L’indagine si concentra sulle perifrasi e . In spagnolo sono presenti due perifrasi formalmente simili ( e ) ma con una gamma di usi più ampia. Sulla base di una rassegna di studi pubblicati su questo tema, si evidenzia che il disapprendimento delle funzioni non condivise è possibile, ma richiede un livello avanzato di competenza e una prolungata esposizione alla L2. I risultati suggeriscono l’importanza di interventi didattici mirati su base contrastiva per facilitare l’acquisizione consapevole delle restrizioni d’uso nella L2, specialmente in contesti di esposizione linguistica limitata
Bioengineering for peripheral arterial diseases: From nanosystems for vascular drug delivery to bioactive small diameter vascular prostheses
No full textMultiple Latent Variable Generative Models: an Information-Theoretic Perspective and Applications
No full textIn recent years, the field of image generation has made significant advancements, as modern deep learning systems became capable of producing photo-realistic pictures of stunning quality. In the vast amount of possible architectures that can perform such a task, we recognize an emerging subclass, which we term Multiple Latent Variable Generative Models (MLVGMs). These systems, developed as an improvement over traditional latent variable frameworks such as Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs), employ multiple latent variables that enter the generative process at subsequent stages, gradually refining image features from coarse, global aspects to finer, local details. Nevertheless, these properties have so far been observed only at an empirical level, and independently for various models.
In this work, we formally recognize MLVGMs as a distinct category of generative models, and propose the first theoretical interpretation that discusses the reasons behind their "global-to-local" subdivision of image information. More specifically, our study is grounded on Information Theory, drawing connections between the generative process of MLVGMs and the framework known as Successive Refinement of Information.
Furthermore, to better understand how the gradual refinement of the image features operates on real models, we develop an algorithm to quantitatively measure this phenomenon, estimating the contribution of each latent variable from the Mutual Information shift produced on real images. The proposed procedure enhances the control over the entire generative process, allowing to use MLVGMs in new and unexplored ways.
On this basis, we study different potential applications of MLVGMs beyond powerful image generation, specifically in tasks such as Self-Supervised Contrastive Representation Learning (SSCRL) and Adversarial Robustness. In SSCRL, we leverage the generative process to sample better positive views, which lead to more discriminative feature learning. In the context of Adversarial Robustness, we employ MLVGMs as foundation models to purify adversarial examples, offering a novel defense mechanism.
Finally, we investigate how our findings can be extended from continuous to discrete variables, allowing the consideration of other types of generative models, and in particular those operating with Vector Quantized Variational Autoencoders (VQ-VAEs). We find that the Successive Refinement framework well suits also this kind of architecture, and propose a new variant that shows some interesting results, broadening the impact of our study
Experimental Characterization of a Bladeless Air Compressor
No full textThe Tesla compressor is an innovative technology that offers a unique approach to fluid compression. Unlike traditional compressors that use rotating blades, bladeless compressors utilize closely spaced disks to create compression. The purpose of this article is to design a prototype Tesla air compressor with optimal design parameters and investigate the performance and loss characteristics based on numerical analysis and experimental demonstration. The prototype model has been numerically investigated at different rotational speeds, and the results have been compared with those obtained in experiments. Computational fluid dynamics (CFD) simulations indicate that the rotor-only efficiency is greater than 90% at very low mass flowrates, while the coupling of the rotor and volute leads to a total-to-static efficiency of approximately 58% (without losses) at 14 g/s. At a nominal mass flow of 4 g/s, the highest total-to-static pressure ratio would be around 1.27. Experimental results indicate leakage losses greatly reduce net mass flow, while pressure ratio values are in good agreement with CFD predictions. During this experiment, a maximum isentropic efficiency of 32.4% is measured. Indeed, the prototype included ventilation and leakage losses, which were not modeled in the CFD analysis. It is remarkable that the compressor does not show any unstable behavior down to zero mass flow (closed valve test), where the CFD and the experiment show consistent pressure ratios. An estimation of the losses from end-wall friction and leakage flow is carried out using numerical simulations at different exit radial clearances. Increasing radial clearance results in an increase in leakage and end-wall power loss, the latter being driven mainly by the axial clearance with the casing, which remained unchanged. To minimize leakage, a Teflon ring has been used as a first measure. Numerical calculations have indicated that the leakage rate is approximately 6 g/s at design speed. A brush seal-type solution can improve the sealing system to reduce leakage
Initial assessments of topographic variability effects on urban pluvial flood modelling accuracy
No full textPluvial flooding is a growing concern due to climate change and urbanization. Accurate flood modeling requires high-resolution topographic data, but data availability, inconsistencies, and computational constraints pose challenges. This study examines the sensitivity of pluvial flood models to digital terrain model (DEM) resolution using HEC-RAS 2D in a frequently flooded Genoa neighborhood. Simulations with 5m, 1m, and 0.5m resolutions assess flood extent, depth, and computational efficiency. Results highlight the importance of high-resolution data for accurate predictions. The research informs urban flood risk management by balancing data granularity with resource efficiency, advocating for microtopographic integration and DEM-DSM comparisons
Assessing the sensitivity of pluvial flood modelling to the topographic description of Urban Areas
No full textGreen Innovation and Environmental Performance: The Moderating Roles of Governance and Policy
No full textDeterminants of response and molecular dynamics in HER2+ER+ breast cancers from the NA-PHER2 trial receiving HER2-targeted and endocrine therapies
No full text: Improved outcomes in HER2+ female breast cancer have resulted from chemotherapy and anti-HER2 therapies. However, HER2+ER+ cancers exhibit lower response rates. The phase 2 NA-PHER2 trial (NCT02530424) investigated chemo-free preoperative HER2 blockade (trastuzumab + pertuzumab) and CDK4/6 inhibition (palbociclib) with or without endocrine therapy (fulvestrant) in HER2+ER+ breast cancer. Clinical endpoints (i.e. Ki67 dynamics and pathological complete response) were previously reported. Here we report on the biomarker analysis, secondary objective of the study. Through RNA sequencing and tumour infiltrating lymphocytes (TIL) assessment in serial biopsies, we identified biomarkers predictive of pCR or Day14 Ki67 response and unveiled treatment-induced molecular changes. High immune infiltration and low ER signalling correlated with pCR, while TP53 mutations associated with high Day14 Ki67. Stratification based on Ki67 at Day14 and at surgery defined three response groups (Ki67 HighHigh, LowHigh, LowLow), with divergent tumour and stroma expression dynamics. The HighHigh group showed dysfunctional immune infiltration and overexpression of therapeutic targets like PAK4 at baseline. The LowLow group exhibited a Luminal A phenotype by the end of treatment. This study expands our understanding of drivers and dynamics of HER2+ER+ tumour response, towards treatment tailoring
Integrating satellite remote sensing and proximal data to investigate the role of brittle tectonics in the distribution of geothermal surface manifestations. Insights from the Parco Naturalistico delle Biancane - Larderello geothermal field (Southern Tuscany, Italy)
No full textGeothermal surface manifestations (e.g., fumaroles, hot springs, geysers, mud-pots, mineral alterations) and thermal anomalies are some of the main indicators of a potential geothermal field. They can be detected by Thermal Infrared (TIR) Remote Sensing satellite data (e.g., Landsat 8 and ECOSTRESS). In this work, we propose a methodology involving image processing techniques combined with field data to investigate the distribution of thermal anomalies and their relationship with the morpho-structural domains related to the tectonic framework in the “Parco Naturalistico delle Biancane” (PNB) area in Southern Tuscany (Italy).
A multiscale approach was applied, analyzing spatial distributions of geothermal domains in relation to structural elements, in order to contribute to understanding how tectonics controls the circulation and geothermal fluid upwelling. An automatic lineament domain analysis has been applied to Digital Elevation Models (DEM), Land Surface Temperature (LST) maps, and high-resolution FLIR thermal images from drones. The data allowed analysis at different spatial resolutions (SR): DEM and LST (30 m resolution) for regional-scale comparisons of structural and thermal fields, and FLIR thermal maps (0.25 m resolution) to explore local-scale relationships between structural features and thermal anomalies observed in the field. Surface temperature maps, derived from satellite and drone data, highlight thermal lineaments and thermal fractures parallel to SW-NE and NW-SE structural systems, which control the upwelling of geothermal fluids. These thermal anomalies are bounded by an N-S structural system (i.e. faults). The study demonstrates that remote sensing techniques can effectively derive heat flow and thermal lineament maps, reflecting fluid circulation contributions. This methodology provides critical insights into the tectonic controls on geothermal systems, offering valuable tools for geothermal resource exploration and assessment. Moreover, it demonstrates the potential of upcoming TIR satellite missions—such as NASA-ASI Surface Biology and Geology SBG-TIR, ISRO–CNES TRISHNA, USGS-NASA Landsat Next, and ESA’s LSTM mission—to enhance the detection and characterization of geothermal features