University of Padua

Padua@thesis
Not a member yet
    41199 research outputs found

    Algoritmi di approssimazione per problemi di bin packing e set covering

    No full text
    [1] Korte B., Vygen J., (2011), Ottimizzazione Combinatoria, Springer, capitoli 16, 18. [2] Vazirani V. V., (2001), Approximation Algorithm, Springer, capitoli 2, 12, 14, 15. [3] Eisenbrand F., Kakimura N., Rothvoß T., SanitàL., (2011), Set Covering with Ordered Replacement: Additive and Multiplicative Gaps, IPCO 2011

    Introduzione alle varietà algebriche

    No full text
    La geometria algebrica è un ramo della matematica che si occupa, nella sua declinazione più elementare, dello studio degli insiemi che possono essere descritti come luogo dei punti di annullamento simultaneo di un insieme di polinomi negli spazi affini e proiettivi. In questo lavoro vengono presentati i concetti di base della geometria algebrica insieme ad alcuni risultati di algebra commutativa, necessari per sviluppare un approccio formale allo studio della geometria. Lo spirito di questa tesi è volto ad uno studio elementare della geometria algebrica, che culmina nella presentazione di alcuni esempi classici come la Superficie di Veronese e l'immersione di Segre

    Towards explainability in knowledge enhanced neural networks

    Get PDF
    Research on Deep Learning has achieved remarkable results in recent years, mainly thanks to the computing power of modern computers and the increasing availability of large data sets. However, deep neural models are universally considered black boxes: they employ sub-symbolic representations of knowledge, which are inherently opaque to human beings trying to derive explanations. In this work, we first give a survey on the research field of Explainable AI, providing more rigorous definitions of the concepts of interpretability and explainability. We then delve deeper in the research field of Neural Symbolic Integration, which tackles the task of integrating the statistical learning power of machine learning with the symbolic and abstract world of logic. Specifically, we analyze Knowledge Enhanced Neural Networks, a special kind of residual layer for neural architectures which makes it possible to inject symbolic logical knowledge inside a neural network. We describe and analyze experimental results on relational data on the task of collective classification, and study how KENN is able to automatically learn the importance of logical rules from the training data. We finally review explainability methods for KENN, proposing ways to extract explanations for the predictions provided by the model

    Caratterizzazione delle sfere metriche nello spazio iperbolico complesso

    No full text
    Lo scopo di questo lavoro è dimostrare un teorema inedito nella letteratura, che fornisce una caratterizzazione delle superfici sferiche nello spazio iperbolico complesso in analogia all'equazione AN(A)=A0A-N(A)=A_{0} per la sfera unitaria immersa in Rn\mathbb{R}^{n} di centro A0A_{0} e direzione normale N(A)N(A)

    Una partita di basket come passeggiata aleatoria

    No full text

    Grandi deviazioni applicate alla Legge di Taylor in ambienti Markoviani

    No full text
    La Legge di Taylor (TL) è una legge empirica e dichiara che la varianza di una variabile aleatoria non negativa è una funzione potenza della sua media. La TL è stata ampiamente verificata in ecologia e l'esponente osservato della media si attesta tipicamente intorno al valore 2. Nella tesi, ci occuperemo di studiare il comportamento di tale esponente nel caso di un modello matematico di crescita moltiplicativo in un ambiente Markoviano. Mostreremo che la causa per cui l'esponente osservato assuma spesso valori vicino a 2 potrebbe derivare dal sottocampionamento. Inoltre, dimostreremo come il valore dell'esponente possa essere utile per cogliere informazioni ecologicamente importanti riguardo alla popolazione sottostante

    Analytical and numerical modelling of self-heating phenomena due to damage in composite structures

    Get PDF
    This work is intended to investigate analytically and numerically the correlation between self-heating phenomena and damage in CFRP composites subjected to cyclic loading. The aim of this work is to elaborate a model that takes into account the material degradation in order to quantify damage evolution. To accomplish this result a new model is set. The idea is to separate and quantify the heat generation sources that are involved during a fatigue test

    Design and genetic algorithm optimization of a PM motor for the Formula SAE race car

    Get PDF
    The electric racecar of the University of Padua, which participates in engineering students competitions, is propelled by 4 electric motors.This thesis deal with the design and optimization of a new PM motor. The torque objectives are defined analysing the drive cycle. The genetic algorithm optimization is computed using the analytical model of the IPM motor. This model uses the lumped parameter magnetic circuit, the slot-opening concentrated conductors and a fictitious saturation coefficient

    Impact analysis of Photovoltaic Distributed Generation and Plug-In Electric Vehicles in a LV distribution network through the evaluation and application of the Non-Synthetic European LV Test System

    Get PDF
    In this work, an altered version of the Non-Synthetic European Low Voltage Test System, from Koirala et al.) is built. Distributed PV generation and PEV representative loads are implemented and simulated in a daily time series power flow analysis for different penetration levels.The results obtained from the stress test comply with the assertions of prior studies, with some exceptions. To accommodate a high PV penetration, the implementation of coordinated control in the grid is mandatory

    15,643

    full texts

    41,199

    metadata records
    Updated in last 30 days.
    Padua@thesis
    Access Repository Dashboard
    Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇