Archivio Istituzionale della Ricerca- Università del Salento
Not a member yet
80217 research outputs found
Sort by
Random features and polynomial rules
Get PDFRandom features models play a distinguished role in the theory of deep learning, describing the behavior of neural networks close to their infinite-width limit. In this work, we present a thorough analysis of the generalization performance of random features models for generic supervised learning problems with Gaussian data. Our approach, built with tools from the statistical mechanics of disordered systems, maps the random features model to an equivalent polynomial model, and allows us to plot average generalization curves as functions of the two main control parameters of the problem: the number of random features N and the size P of the training set, both assumed to scale as powers in the input dimension D. Our results extend the case of proportional scaling between N, P and D. They are in accordance with rigorous bounds known for certain particular learning tasks and are in quantitative agreement with numerical experiments performed over many order of magnitudes of N and P. We find good agreement also far finite
La pianificazione standardizzata e il «caso Milano»: riflessioni dopo un’audizione parlamentare
Get PDFAnisotropic local covariance matrices for spatial blind source separation
Get PDFThis paper aims to introduce a novel approach to spatial blind source separation (SBSS) that addresses the limitations of existing methods. Current SBSS techniques rely on the joint diagonalization of multiple local covariance functions, all of which assume isotropy. To overcome this constraint, anisotropic local covariance matrices that relax the isotropy assumption are proposed. A simulation study and an appli- cation on real-world data demonstrate the performance improvement obtained by incorporating these anisotropic covariance matrices into the SBSS framework and highlight the potential of this new approach for more accurate and flexible source separation in spatial data analysis
Evidence of nanoCaCO3@PAE internalization in tobacco
No full textNanoparticles hold great potential as innovative agrochemicals, offering significant advantages over traditional products like insecticides, herbicides and fertilizers, which often harm soil and human health. This study investigates the effects of free aqueous pomegranate extract (PAE) and PAE encapsulated in CaCO3 nanocrystals (nanoCaCO3@PAE) on cellular uptake and internalization mechanisms in tobacco plants. Our experiments assessed the cytotoxic effect of free PAE on protoplast viability, while the negative effect was reduced in the presence of nanoCaCO3@PAE. FM4-64 labelling and secRFP-HDEL-transformed protoplasts showed an indirect increased of endocytic activity when treated with 0.25 mg/mL nanoCaCO3@PAE. After optimization of confocal microscope setting, auto-fluorescence of nano-encapsulated extracts were visible in protoplast endosome-like structures. In 3-month-old tobacco seedlings, foliar spraying of PAE or nanoCaCO3@PAE showed no visible signs of stress or necrosis 14 days after treatments, as well as chlorophyll and carotenoid content, compared to control seedlings. On the other hand, a significant increase in the amount of soluble phenols in the treated leaves, particularly in those treated with nanoCaCO3@PAE compared to their free form was observed. The use of fluorescent carbon nanocrystals, nanoCaCO3@FITC, allowed the visualization of the nanoparticles into their final destination, i.e. central vacuole, where the phenols extracted from leaves were not degraded and then probably released from nanoCaCO3@PAE due to the acidic nature of this compartment. Through the use of Tyrphostin A23, a specific inhibitor of clathrin-mediated internalization, it has been demonstrated that nanoCaCO3@FITC were internalized through clathrin-independent mechanisms
Un fiume in piena. Potenzialità e fragilità nell'alunno plusdotato. Quadro epistemologico e risposte adattive, F. Baccassino, Lecce, Pensa Multimedia, 2024
No full textD.A.N.T.E.-U: Design and development of a UDL-based platform for University faculty training
No full textDifferential transcript and soluble factor patterns in macrophage/enterocyte-like monolayer co-cultures based on apical or basolateral LPS exposure
No full textBackground: The monolayer of intestinal epithelial cells (IECs) plays a crucial role in controlling intestinal homeostasis, also by its interaction with the immune system, via paracrine cytokine production, thus driving innate responses by tissue-resident immune cells. Here, using a co-culture model, we investigated the interactions between differentiated Caco-2 cells in monolayer and macrophages, by mimicking the cross-talk between enterocytes and immune cells during gastrointestinal (GI) tract inflammation. Methods: Caco-2 mature monolayers grown on Transwell membranes were challenged with apical or basolateral LPS. After stimulations, the enterocyte-like monolayers were transferred in co-culture with THP-1 derived macrophages. The functional impact of treatments was evaluated in terms of monolayer’s permeability, expression of mRNAs related to inflammation and immune responses and analysis of immune soluble factors present in the co-culture media. Results: LPS effectively affected monolayer’s permeability and induced a pro-inflammatory transcriptional program in Caco-2 monolayers. Remarkably, THP-1 derived macrophages differentially responded based on the diverse directional source of LPS, previously administered to the Caco-2 monolayers. Basolateral sensing of LPS, by Caco-2 monolayers, induced specific increase of several pro-inflammatory factors such as NF-kB1, IL-6 and IL-8, at transcript level, in macrophages, while apical sensing triggering targeted increase of IL-1β expression. Significantly, the analysis of immune factors secreted in the co-culture media suggested that paracrine interactions between enterocyte-like monolayers and macrophages are differently driven based on the basolateral vs. apical inflammation, previously triggered by LPS against the epithelial monolayer, and thus involving different immune gene networks. Conclusions: Taken together, our results suggest a framework of interactions between IECs and macrophages, depending upon the “polarized” inflammatory dysregulation
Enhancing Data Ingestion Efficiency in Cloud-Based Systems: A Design Pattern Approach
Get PDFThis paper aims to defne design patterns specifcally for data ingestion techniques within cloud-based architectures, addressing the challenges associated with high-volume data processing. The approach utilizes a fexible, metadata-driven framework
that enhances adaptability and ease of use. This framework supports both incremental and full refresh methods, allowing
for seamless changes to ingestion types, schema updates, table additions, and the incorporation of new data sources with
minimal intervention from data engineers. The proposed design patterns were validated through experiments conducted on
the Azure and Google Cloud platforms. The experiments demonstrate that the proposed design patterns signifcantly reduce
data ingestion time, showcasing their efectiveness in managing high-volume data ingestion. This paper contributes to the
feld of data management by presenting a comprehensive defnition of design patterns tailored for data ingestion in cloudbased architectures, efectively addressing key challenges in high-volume data processing