The retrieval of geo-/bio-physical variables from remote sensing imagery is a challenging and important research field. On the one hand, advances in electronics, engineering and space sciences are offering to the users community new sensors capable to acquire information on the Earth surface with higher accuracy and improved features with respect to the past. On the other hand, the need of large-scale, accurate and up-to-date mapping and monitoring of natural targets and physical processes is becoming fundamental for many application domains. This calls for the development of accurate, robust and effective retrieval methodologies. The main goal of this thesis is to investigate and develop advanced methods and systems for the retrieval of geo-/bio-physical variables from satellite remote sensing imagery being able to exploit the potential of new and upcoming satellite systems and support real application domains. Special attention has been devoted to the definition of methods and to the analysis of data acquired in the challenging mountain environment. The activity carried out and presented in this dissertation is oriented to investigate the main limitations of the existing methodologies for addressing the estimation problem and to develop novel and improved systems that can overcome the drawbacks identified. In particular, the following main novel contributions are proposed in this thesis: a) A theoretical and empirical comparative analysis of non-linear machine learning regression methods, namely the Multi-Layer Perceptron Neural Network and the Support Vector Regression, for soil moisture retrieval in different operational scenarios. b) A novel multi-objective model-selection strategy for tuning the free parameters of non-linear regression methods taking into account different quality metrics that are jointly optimized. c) A novel hybrid approach to the retrieval of geo-/bio-physical variables from remote sensing data integrating both theoretical electromagnetic models and field reference measurements. d) A sensitivity analysis and a retrieval system for soil moisture content estimation from new generation SAR imagery in an Alpine catchment. e) An empirical study on the effectiveness of fully-polarimetric SAR signals for soil moisture estimation in mountain areas. f) An improved algorithm for mapping and monitoring Green Area Index (GAI) in Alpine pastures and meadows from satellite MODIS imagery. Qualitative and quantitative experimental results obtained on real remotely sensed data confirm the effectiveness of the proposed solutions.

Deep learning is an emerging technology that is considered one of the most promising directions for reaching higher levels of artificial intelligence. Among the other achievements, building computers that understand speech represents a crucial leap towards intelligent machines. Despite the great efforts of the past decades, however, a natural and robust human-machine speech interaction still appears to be out of reach, especially when users interact with a distant microphone in noisy and reverberant environments. The latter disturbances severely hamper the intelligibility of a speech signal, making Distant Speech Recognition (DSR) one of the major open challenges in the field. This thesis addresses the latter scenario and proposes some novel techniques, architectures, and algorithms to improve the robustness of distant-talking acoustic models. We first elaborate on methodologies for realistic data contamination, with a particular emphasis on DNN training with simulated data. We then investigate on approaches for better exploiting speech contexts, proposing some original methodologies for both feed-forward and recurrent neural networks. Lastly, inspired by the idea that cooperation across different DNNs could be the key for counteracting the harmful effects of noise and reverberation, we propose a novel deep learning paradigm called “network of deep neural networks”. The analysis of the original concepts were based on extensive experimental validations conducted on both real and simulated data, considering different corpora, microphone configurations, environments, noisy conditions, and ASR tasks.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

Cellular materials are characterized by a complex interconnected structure of struts or plates and shells which make up the cells edges and faces. Their structure can be advantageously engineered in order to tailor their properties according to the specific application. This aspect makes them particularly attractive for the manufacturing of bone prosthetics since, compared to traditional fully dense implants, although more complex to produce and with less predictable properties, implants with a highly porous structure can be manufactured to match the bone stiffness and at the same time favor bone ingrowth and regeneration. The development of Selective Laser Melting (SLM) made possible to obtain metallic cellular materials with highly complex structures characterized by a wide range of cell morphologies that allow to finely tune the mechanical properties of the implant to the patient needs. Titanium alloys such as Ti-6Al-4V have shown excellent biocompatibility combined with good mechanical properties and have also been successfully used in the manufacturing of lattice structures with minute details via SLM. Nevertheless, there are still several issues to consider. For instance, despite the static mechanical properties of such lattices being addressed by many studies, the fatigue behavior still remains little investigated, even though it is a critical aspect in load-bearing biomedical implants (consider, for example, the periodic nature of human gait in the case of hip implants). In this regard, increasing the fatigue resistance of cellular lattices by finely adjusting the geometry, for instance by adding fillets at the cell-wall joints, is a new interesting opportunity made possible by additive manufacturing technologies. On the other hand, a discrepancy between the as-designed and the as-built geometry in SLM parts is an issue that can be critically important for lattices with pore size and strut thicknesses of a few hundred microns, such as biomedical lattices. Indeed, any geometrical imperfection introduces a degree of uncertainty that can alter the mechanical properties of the as-built lattice. This work represents an attempt in the direction of building a deeper understanding of the effect of the fine geometrical details, such as the fillet radius at the joints and the thickness of the struts, on the elastic constants and on the fatigue resistance of Ti-6Al-4V SLM lattices, with the aim to develop analytical predictive models of the mechanical properties. Moreover, this work also aims at investigating the as-built/as-designed morphological discrepancy in lattices in relation to the their as-designed geometry and its effects on the elastic modulus and the fatigue resistance. In this regard, the purpose is to develop quantitative relationships between the as-designed and the as-built geometry in order to obtain design tools to predict the final morphology of the lattice by taking into account the manufacturing errors. This thesis covers a wide range of topics, therefore, in the interest of a better presentation, the results of the research have been devided into three independent Chapters. Each of them has been provided of an abstract and an Introduction and divided into a Materials and Methods (or Modelling) section, a Results and Discussion section and finally Conclusions and References. Naturally, the chapters are logically connected and coherent with the frame defined by the title of the thesis. Therefore, this thesis is organized into five chapters. In the first Chapter the backrground to the topics discussed in the subsequent chapters is provided and the relevant literature is reviewed, while in the fifth and last Chapter some conclusions are drawn, and future perspectives are discussed. The core of the work is contained in the three central chapters. In Chapter II, analytical models developed to predict the elastic constants and the stress concentration factors (SCF) of 2D lattices with variously arranged square cells and filleted junctions are presented. The effect of stretching and bending actions on the elastic constants of a single cell is identified by devising an analytical model based on classical beam theory and and periodic boundary conditions. Specifically, two spatial arrangements are considered: a honeycomb with regular square cells and a honeycomb with square cells staggered by a prescribed offset of half of the cell wall length. The theoretical beam model is fitted to the results of a 2D Finite Elements (FE) model based on plane elements via an extensive parametric analysis. In this way, semi-analytical formulas are proposed to calculate the stiffness in large domains of the geometric parameters (strut thickness t0 and fillet radius R). A numerical method is also proposed to estimate the SCFs at the cell wall junctions of a 2D regular square cellular lattice. The aim is to obtain a model capable of calculating the values of the SCF as a function of the unit cell geometrical parameters and consequently assess the stress state in the lattice, which is one of the main factors determining fatigue resistance. This was achieved by applying the FE method to the unit cell for wide intervals of t0 and R to calculate the SCF for each couple of the parameters. The values of the SCFs were then fitted with functions. The models developed in this Chapter are then used in the subsequent chapters as a support in the design of 3D regular square lattices and in the interpretation of the mechanical characterization. In Chapter III, the results of the mechanical and morphological characterization of different regular cubic open-cell cellular structures produced via SLM of Ti-6Al-4V alloy, all with the same nominal elastic modulus of 3 GPa that matches that of human trabecular bone, are presented. The fully reversed fatigue strength at 106 cycles and the elastic modulus were measured and an attempt was made to link them to the manufacturing defects (porosity and geometrical inaccuracies). Half of the specimens was subjected to a stress relief thermal treatment while the other half to Hot Isostatic Pressing (HIP), and the effect of the treatments on porosity and on the mechanical properties was assessed. The results of fatigue and quasi-static tests on regular cubic lattices were compared with FE calculations based on the as-designed geometry and on the as-built geometry reconstructed from micro X-ray computed tomography (ÂμCT) scans. It was observed that the fatigue strength and, to a lesser extent, the elastic modulus are correlated with the number and severity of defects and that predictions on the mechanical properties based on the as-designed geometry are not accurate. The fatigue strength seems to be highly dependent on the surface irregularities and on the notches introduced during the manufacturing process. In fully reversed fatigue tests, the high performances of stretching dominated structures compared to bending dominated structures are not found. In fact, with thicker struts, such structures proved to be more resistant, even if bending actions were present. Given the small size of the unit cells (the unit cell size is 1.5 mm and the strut thickness is 0.26 mm) and the limitations in accuracy of the printer, the fillet radii at the junctions were highly irregular and somewhat hard to recognize. In order to investigate the real benefit of filleted junctions on the stress concentration effects at the junctions and to assess the manufacturability of such minute geometrical detail, a new experimental campaign was set up. In Chapter IV, a set of cubic lattice specimens with filleted junctions was designed and produced via SLM. The size of the unit cell is considerably larger than that of the previous specimens, being 8 mm, 6 mm and 4 mm with the rest of the geometrical parameters scaled accordingly. Thus, nine combinations of the geometrical parameters of the unit cell and three orientations with respect to the printing direction are considered. The aim is to investigate the relationship between the as-designed and the as-built geometry and to find the smallest radius which can be accurately reproduced by the printer. Moreover, a compensation strategy of the morphological defects is devised using the mathematical relationships obtained between the as-designed and the as-built strut thickness. This strategy consists in modifying the input CAD to compensate for the deviations introduced by the SLM process.

Endohedral metallofullerenes (EMFs) have attracted increasing attention during past decades due to their novel structures and potential applications in a variety of fields such as biomedical applications and molecular electronics. This dissertation addresses the structural characterization and hydrogenation of EMFs. A family of novel large cage yttrium-based TNT EMFs Y₃N@C₂ₙ (n=40-44) was prepared, separated, and structurally characterized for the first time. The structure of Y₃N@C₂ₙ (n=40-44) is proposed by the experimental and computational ¹³C NMR studies. The first ⁸⁹Y NMR results for Y₃N@<I>Iₕ</i>-C₈₀, Y₃N@<I>Cₛ</i>-C₈₄ and Y₃N@<I>D₃</i>-C₈₆ reveal a progression from isotropic to restricted (Y₃N)⁶⁺</sup> cluster motional processes. The di-metallic EMF Y₂C₉₄ is distinguished as a metal-carbide based EMF, Y₂C₂@<I>D₃</i>-C₉₂. The carbide within the cage is successfully detected by ¹³C NMR. The scalar J_Y-C coupling between the yttrium atoms and the C₂ unit within the C₉₂ cage is successfully observed, suggesting the C₂ unit rotates rapidly around the yttrium atoms. Two paramagnetic endohedral metalloheterofullerenes, Y₂@C₇₉N and Gd₂@C₇₉N, were also synthesized and characterized. The EPR study demonstrated that the spin density is mainly localized between the two metallic ions. A spin-site exchange system could be constructed between Y₂@C₇₉N and the organic donor TMPD. Being a unique paramagnetic material, Gd₂@C₇₉N displays an unusual stability over a wide temperature range, which could be very useful in optical and magnetic areas. Functionalization of EMFs is another point of interest in this dissertation. Hydrogenated Sc₃N@C₈₀ was synthesized and characterized. Our study demonstrated that the Sc₃N@C₈₀ can be fully hydrogenated and the pristine Sc₃N@C₈₀ can be recovered from Sc₃N@C₈₀H₈₀ after being heated in vacuum. The hydrogenated EMFs could be potential hydrogen storage materials. / Ph. D.