Geometric deep learning for polygonal mesh denoising

Il seguente lavoro si propone come analisi degli operatori convoluzionali che caratterizzano le graph neural networks. ln particolare, la trattazione si divide in due parti, una teorica e una sperimentale. Nella parte teorica vengono innanzitutto introdotte le nozioni preliminari di mesh e convoluzione su mesh. In seguito vengono riportati i concetti base del geometric deep learning, quali le definizioni degli operatori convoluzionali e di pooling e unpooling. Un'attenzione particolare è stata data all'architettura Graph U-Net. La parte sperimentare riguarda l'applicazione delle reti neurali e l'analisi degli operatori convoluzionali applicati al denoising di superfici perturbate a causa di misurazioni imperfette effettuate da scanner 3D.

Sul rinforzo di elementi in C.A.: confronto tra varie normative

Il documento pre-normativo italiano sul rinforzo di strutture in c.a. mediante l’uso di materiale fibrorinforzato. 1.1 INTRODUZIONE La situazione unica dell’Italia per quanto riguarda la conservazione delle costruzioni esistenti, è il risultato della combinazione di due aspetti, come primo, il medio-alto rischio sismico di una gran parte di territorio, come testimoniato dalla zonizzazione sismica recente, e come secondo aspetto, l'estrema complessità di un ambiente edilizio che non ha confronto nel mondo. Le tipologie della costruzione in Italia si distinguono a quelle stimate come patrimonio storico, che in alcuni casi risalgono a circa 2000 anni fa, a quelle che sono state costruite in ultimi cinque secoli, durante e dopo il Rinascimento, che sono considerate come patrimonio culturale ed architettonico dell' Italia (e del mondo!), infine a quelle fatte in tempi recenti, considerevolmente durante e dopo il boom economico del l960 ed ora visti come antiquate. Le due prime categorie in gran parte sono composte dalle edilizie di muratura, mentre agli ultimi principalmente appartengono le costruzioni di cemento armato.

Power consumption analysis for self-diagnosis of Wireless Sensor Nodes

Wireless sensor networks can transform our buildings in smart environments, improving comfort, energy efficiency and safety. Today however, wireless sensor networks are not considered reliable enough for being deployed on large scale. In this thesis, we study the main failure causes for wireless sensor networks, the existing solutions to improve reliability and investigate the possibility to implement self-diagnosis through power consumption measurements on the sensor nodes. Especially, we focus our interest on faults that generate in-range errors: those are wrong readings but belong to the range of the sensor and can therefore be missed by external observers. Using a wireless sensor network deployed in the R\&D building of NXP at the High Tech Campus of Eindhoven, we performed a power consumption characterization of the Wireless Autonomous Sensor (WAS), and studied through some experiments the effect that faults have in the power consumption of the sensor.

Analisi di morfologie carsiche o crio-carsiche in relazione alla mobilizzazione di fluidi nella regione di Arabia Terra, Marte.

This thesis tries to interpret the origin and evolution of karst-like forms present in Arabia Terra, a region of Mars that develops in the equatorial zone of the planet. The work has been carried out specifically in the craters Crommelin (4o 91’ N-10o 51’ E), 12000088 (3o 48’ N-1o 30’ E), NE 12000088 (4° 20’ N-2° 50’ E), C "2" (3° 54’ N-1° W), and in their surrounding areas. These craters contain layered deposits characterized by a high albedo and on which erosion is very pronounced. The area containing the craters is a plateau that has the same characteristics of albedo and texture. The preliminary morphological study has made use of instrumentation such as the Mars Reconnaissance Orbiter (MRO), in particular HiRISE images (High Resolution Imaging Science Experiment), CTX (Context Camera) and CRISM (Compact Reconnaissance Imaging Spectrometers for Mars). A regional geomorphological map has been drawn up containing the main morphotypes, and detailed geomorphological maps were prepared for different karst-like morphologies. The analysis of spectral data collected from CRISM instrumentation has allowed to identify the footprint of sulphate minerals in the external area. Data were collected for morphometric negative forms (karst-like) and positive forms (mud volcanoes, dikes and pingos). For the analysis of the relief forms DTMs (Digital Terrain Models) produced by the union of stereographic CTX couples or HiRISE were used. From the analysis of high-resolution images morphological footprints similar to periglacial environments have been identified, including the presence of patterned ground and polygonal cracks found all over the area of investigation, and relief structures similar to pingos present in the crater C "2". These observations allow us to imagine a geological past with a cold climate at the equator able to freeze the few fluids present in the Martian arid terrain. The development of karst-like landforms, on the other hand, can be attributed to a subsequent improval of the weather conditions that led to a normal climate regime for the equatorial areas, resulting in the degradation of the permafrost. The melt waters have thus allowed the partial dissolution of the sulphate layers. The karst-like forms look rather fresh suggesting them to be not that old.