Data driven regularization models of non-linear ill-posed inverse problems in imaging

Imaging technologies are widely used in application fields such as natural sciences, engineering, medicine, and life sciences. A broad class of imaging problems reduces to solve ill-posed inverse problems (IPs). Traditional strategies to solve these ill-posed IPs rely on variational regularization methods, which are based on minimization of suitable energies, and make use of knowledge about the image formation model (forward operator) and prior knowledge on the solution, but lack in incorporating knowledge directly from data. On the other hand, the more recent learned approaches can easily learn the intricate statistics of images depending on a large set of data, but do not have a systematic method for incorporating prior knowledge about the image formation model. The main purpose of this thesis is to discuss data-driven image reconstruction methods which combine the benefits of these two different reconstruction strategies for the solution of highly nonlinear ill-posed inverse problems. Mathematical formulation and numerical approaches for image IPs, including linear as well as strongly nonlinear problems are described. More specifically we address the Electrical impedance Tomography (EIT) reconstruction problem by unrolling the regularized Gauss-Newton method and integrating the regularization learned by a data-adaptive neural network. Furthermore we investigate the solution of non-linear ill-posed IPs introducing a deep-PnP framework that integrates the graph convolutional denoiser into the proximal Gauss-Newton method with a practical application to the EIT, a recently introduced promising imaging technique. Efficient algorithms are then applied to the solution of the limited electrods problem in EIT, combining compressive sensing techniques and deep learning strategies. Finally, a transformer-based neural network architecture is adapted to restore the noisy solution of the Computed Tomography problem recovered using the filtered back-projection method.

Iterative regularization methods for ill-posed problems

This Ph.D thesis focuses on iterative regularization methods for regularizing linear and nonlinear ill-posed problems. Regarding linear problems, three new stopping rules for the Conjugate Gradient method applied to the normal equations are proposed and tested in many numerical simulations, including some tomographic images reconstruction problems. Regarding nonlinear problems, convergence and convergence rate results are provided for a Newton-type method with a modified version of Landweber iteration as an inner iteration in a Banach space setting.

Regularization meets GreenAI: a new framework for image reconstruction in life sciences applications

Ill-conditioned inverse problems frequently arise in life sciences, particularly in the context of image deblurring and medical image reconstruction. These problems have been addressed through iterative variational algorithms, which regularize the reconstruction by adding prior knowledge about the problem's solution. Despite the theoretical reliability of these methods, their practical utility is constrained by the time required to converge. Recently, the advent of neural networks allowed the development of reconstruction algorithms that can compute highly accurate solutions with minimal time demands. Regrettably, it is well-known that neural networks are sensitive to unexpected noise, and the quality of their reconstructions quickly deteriorates when the input is slightly perturbed. Modern efforts to address this challenge have led to the creation of massive neural network architectures, but this approach is unsustainable from both ecological and economic standpoints. The recently introduced GreenAI paradigm argues that developing sustainable neural network models is essential for practical applications. In this thesis, we aim to bridge the gap between theory and practice by introducing a novel framework that combines the reliability of model-based iterative algorithms with the speed and accuracy of end-to-end neural networks. Additionally, we demonstrate that our framework yields results comparable to state-of-the-art methods while using relatively small, sustainable models. In the first part of this thesis, we discuss the proposed framework from a theoretical perspective. We provide an extension of classical regularization theory, applicable in scenarios where neural networks are employed to solve inverse problems, and we show there exists a trade-off between accuracy and stability. Furthermore, we demonstrate the effectiveness of our methods in common life science-related scenarios. In the second part of the thesis, we initiate an exploration extending the proposed method into the probabilistic domain. We analyze some properties of deep generative models, revealing their potential applicability in addressing ill-posed inverse problems.

Bayesian Analysis of Linear Inverse Problems with Applications in Economics and Finance

In my PhD thesis I propose a Bayesian nonparametric estimation method for structural econometric models where the functional parameter of interest describes the economic agent's behavior. The structural parameter is characterized as the solution of a functional equation, or by using more technical words, as the solution of an inverse problem that can be either ill-posed or well-posed. From a Bayesian point of view, the parameter of interest is a random function and the solution to the inference problem is the posterior distribution of this parameter. A regular version of the posterior distribution in functional spaces is characterized. However, the infinite dimension of the considered spaces causes a problem of non continuity of the solution and then a problem of inconsistency, from a frequentist point of view, of the posterior distribution (i.e. problem of ill-posedness). The contribution of this essay is to propose new methods to deal with this problem of ill-posedness. The first one consists in adopting a Tikhonov regularization scheme in the construction of the posterior distribution so that I end up with a new object that I call regularized posterior distribution and that I guess it is solution of the inverse problem. The second approach consists in specifying a prior distribution on the parameter of interest of the g-prior type. Then, I detect a class of models for which the prior distribution is able to correct for the ill-posedness also in infinite dimensional problems. I study asymptotic properties of these proposed solutions and I prove that, under some regularity condition satisfied by the true value of the parameter of interest, they are consistent in a "frequentist" sense. Once I have set the general theory, I apply my bayesian nonparametric methodology to different estimation problems. First, I apply this estimator to deconvolution and to hazard rate, density and regression estimation. Then, I consider the estimation of an Instrumental Regression that is useful in micro-econometrics when we have to deal with problems of endogeneity. Finally, I develop an application in finance: I get the bayesian estimator for the equilibrium asset pricing functional by using the Euler equation defined in the Lucas'(1978) tree-type models.

Homogenization of nonlinear composites for multiscale analysis

A composite is a material made out of two or more constituents (phases) combined together in order to achieve desirable mechanical or thermal properties. Such innovative materials have been widely used in a large variety of engineering fields in the past decades. The design of a composite structure requires the resolution of a multiscale problem that involves a macroscale (i.e. the structural scale) and a microscale. The latter plays a crucial role in the determination of the material behavior at the macroscale, especially when dealing with constituents characterized by nonlinearities. For this reason, numerical tools are required in order to design composite structures by taking into account of their microstructure. These tools need to provide an accurate yet efficient solution in terms of time and memory requirements, due to the large number of internal variables of the problem. This issue is addressed by different methods that overcome this problem by reducing the number of internal variables. Within this framework, this thesis focuses on the development of a new homogenization technique named Mixed TFA (MxTFA) in order to solve the homogenization problem for nonlinear composites. This technique is based on a mixed-stress variational approach involving self-equilibrated stresses and plastic multiplier as independent variables on the Reference Volume Element (RVE). The MxTFA is developed for the case of elastoplasticity and viscoplasticity, and it is implemented into a multiscale analysis for nonlinear composites. Numerical results show the efficiency of the presented techniques, both at microscale and at macroscale level.

Structural analysis of combinatorial optimization problem characteristics and their resolution using hybrid approaches

Many combinatorial problems coming from the real world may not have a clear and well defined structure, typically being dirtied by side constraints, or being composed of two or more sub-problems, usually not disjoint. Such problems are not suitable to be solved with pure approaches based on a single programming paradigm, because a paradigm that can effectively face a problem characteristic may behave inefficiently when facing other characteristics. In these cases, modelling the problem using different programming techniques, trying to ”take the best” from each technique, can produce solvers that largely dominate pure approaches. We demonstrate the effectiveness of hybridization and we discuss about different hybridization techniques by analyzing two classes of problems with particular structures, exploiting Constraint Programming and Integer Linear Programming solving tools and Algorithm Portfolios and Logic Based Benders Decomposition as integration and hybridization frameworks.

Characterization of geothermal reservoirs’ parameters by inverse problem resolution and geostatistical simulations

BTES (borehole thermal energy storage)systems exchange thermal energy by conduction with the surrounding ground through borehole materials. The spatial variability of the geological properties and the space-time variability of hydrogeological conditions affect the real power rate of heat exchangers and, consequently, the amount of energy extracted from / injected into the ground. For this reason, it is not an easy task to identify the underground thermal properties to use when designing. At the current state of technology, Thermal Response Test (TRT) is the in situ test for the characterization of ground thermal properties with the higher degree of accuracy, but it doesn’t fully solve the problem of characterizing the thermal properties of a shallow geothermal reservoir, simply because it characterizes only the neighborhood of the heat exchanger at hand and only for the test duration. Different analytical and numerical models exist for the characterization of shallow geothermal reservoir, but they are still inadequate and not exhaustive: more sophisticated models must be taken into account and a geostatistical approach is needed to tackle natural variability and estimates uncertainty. The approach adopted for reservoir characterization is the “inverse problem”, typical of oil&gas field analysis. Similarly, we create different realizations of thermal properties by direct sequential simulation and we find the best one fitting real production data (fluid temperature along time). The software used to develop heat production simulation is FEFLOW 5.4 (Finite Element subsurface FLOW system). A geostatistical reservoir model has been set up based on literature thermal properties data and spatial variability hypotheses, and a real TRT has been tested. Then we analyzed and used as well two other codes (SA-Geotherm and FV-Geotherm) which are two implementation of the same numerical model of FEFLOW (Al-Khoury model).

Eco-Management and Audit Scheme (EMAS): An evaluation of success factors in the agricultural sector. The case of Italy's Emilia Romagna region.

Different tools have been used to set up and adopt the model for the fulfillment of the objective of this research. 1. The Model The base model that has been used is the Analytical Hierarchy Process (AHP) adapted with the aim to perform a Benefit Cost Analysis. The AHP developed by Thomas Saaty is a multicriteria decision - making technique which decomposes a complex problem into a hierarchy. It is used to derive ratio scales from both discreet and continuous paired comparisons in multilevel hierarchic structures. These comparisons may be taken from actual measurements or from a fundamental scale that reflects the relative strength of preferences and feelings. 2. Tools and methods 2.1. The Expert Choice Software The software Expert Choice is a tool that allows each operator to easily implement the AHP model in every stage of the problem. 2.2. Personal Interviews to the farms For this research, the farms of the region Emilia Romagna certified EMAS have been detected. Information has been given by EMAS center in Wien. Personal interviews have been carried out to each farm in order to have a complete and realistic judgment of each criteria of the hierarchy. 2.3. Questionnaire A supporting questionnaire has also been delivered and used for the interviews . 3. Elaboration of the data After data collection, the data elaboration has taken place. The software support Expert Choice has been used . 4. Results of the Analysis The result of the figures above (vedere altro documento) gives a series of numbers which are fractions of the unit. This has to be interpreted as the relative contribution of each element to the fulfillment of the relative objective. So calculating the Benefits/costs ratio for each alternative the following will be obtained: Alternative One: Implement EMAS Benefits ratio: 0, 877 Costs ratio: 0, 815 Benfit/Cost ratio: 0,877/0,815=1,08 Alternative Two: Not Implement EMAS Benefits ratio: 0,123 Costs ration: 0,185 Benefit/Cost ratio: 0,123/0,185=0,66 As stated above, the alternative with the highest ratio will be the best solution for the organization. This means that the research carried out and the model implemented suggests that EMAS adoption in the agricultural sector is the best alternative. It has to be noted that the ratio is 1,08 which is a relatively low positive value. This shows the fragility of this conclusion and suggests a careful exam of the benefits and costs for each farm before adopting the scheme. On the other part, the result needs to be taken in consideration by the policy makers in order to enhance their intervention regarding the scheme adoption on the agricultural sector. According to the AHP elaboration of judgments we have the following main considerations on Benefits: - Legal compliance seems to be the most important benefit for the agricultural sector since its rank is 0,471 - The next two most important benefits are Improved internal organization (ranking 0,230) followed by Competitive advantage (ranking 0, 221) mostly due to the sub-element Improved image (ranking 0,743) Finally, even though Incentives are not ranked among the most important elements, the financial ones seem to have been decisive on the decision making process. According to the AHP elaboration of judgments we have the following main considerations on Costs: - External costs seem to be largely more important than the internal ones (ranking 0, 857 over 0,143) suggesting that Emas costs over consultancy and verification remain the biggest obstacle. - The implementation of the EMS is the most challenging element regarding the internal costs (ranking 0,750).

The gaseous environment of radio galaxies: a new perspective from high-resolution x-ray spectroscopy

It is known that massive black holes have a profound effect on the evolution of galaxies, and possibly on their formation by regulating the amount of gas available for the star formation. However, how black hole and galaxies communicate is still an open problem, depending on how much of the energy released interacts with the circumnuclear matter. In the last years, most studies of feedback have primarily focused on AGN jet/cavity systems in the most massive galaxy clusters. This thesis intends to investigate the feedback phenomena in radio--loud AGNs from a different perspective studying isolated radio galaxies, through high-resolution spectroscopy. In particular one NLRG and three BLRG are studied, searching for warm gas, both in emission and absorption, in the soft X-ray band. I show that the soft spectrum of 3C33 originates from gas photoionized by the central engine. I found for the first time WA in 3C382 and 3C390.3. I show that the observed warm emitter/absorbers is not uniform and probably located in the NLR. The detected WA is slow implying a mass outflow rate and kinetic luminosity always well below 1% the L(acc) as well as the P(jet). Finally the radio--loud properties are compared with those of type 1 RQ AGNs. A positive correlation is found between the mass outflow rate/kinetic luminosity, and the radio loudness. This seems to suggest that the presence of a radio source (the jet?) affects the distribution of the absorbing gas. Alternatively, if the gas distribution is similar in Seyferts and radio galaxies, the M(out) vs rl relation could simply indicate a major ejection of matter in the form of wind in powerful radio AGNs.

Local Earthquake Tomography by trans-dimensional Monte Carlo Sampling

In this study a new, fully non-linear, approach to Local Earthquake Tomography is presented. Local Earthquakes Tomography (LET) is a non-linear inversion problem that allows the joint determination of earthquakes parameters and velocity structure from arrival times of waves generated by local sources. Since the early developments of seismic tomography several inversion methods have been developed to solve this problem in a linearized way. In the framework of Monte Carlo sampling, we developed a new code based on the Reversible Jump Markov Chain Monte Carlo sampling method (Rj-McMc). It is a trans-dimensional approach in which the number of unknowns, and thus the model parameterization, is treated as one of the unknowns. I show that our new code allows overcoming major limitations of linearized tomography, opening a new perspective in seismic imaging. Synthetic tests demonstrate that our algorithm is able to produce a robust and reliable tomography without the need to make subjective a-priori assumptions about starting models and parameterization. Moreover it provides a more accurate estimate of uncertainties about the model parameters. Therefore, it is very suitable for investigating the velocity structure in regions that lack of accurate a-priori information. Synthetic tests also reveal that the lack of any regularization constraints allows extracting more information from the observed data and that the velocity structure can be detected also in regions where the density of rays is low and standard linearized codes fails. I also present high-resolution Vp and Vp/Vs models in two widespread investigated regions: the Parkfield segment of the San Andreas Fault (California, USA) and the area around the Alto Tiberina fault (Umbria-Marche, Italy). In both the cases, the models obtained with our code show a substantial improvement in the data fit, if compared with the models obtained from the same data set with the linearized inversion codes.

GNSS interference management techniques against malicious attacks

This thesis collects the outcomes of a Ph.D. course in Telecommunications Engineering and it is focused on the study and design of possible techniques able to counteract interference signal in Global Navigation Satellite System (GNSS) systems. The subject is the jamming threat in navigation systems, that has become a very increasingly important topic in recent years, due to the wide diffusion of GNSS-based civil applications. Detection and mitigation techniques are developed in order to fight out jamming signals, tested in different scenarios and including sophisticated signals. The thesis is organized in two main parts, which deal with management of GNSS intentional counterfeit signals. The first part deals with the interference management, focusing on the intentional interfering signal. In particular, a technique for the detection and localization of the interfering signal level in the GNSS bands in frequency domain has been proposed. In addition, an effective mitigation technique which exploits the periodic characteristics of the common jamming signals reducing interfering effects at the receiver side has been introduced. Moreover, this technique has been also tested in a different and more complicated scenario resulting still effective in mitigation and cancellation of the interfering signal, without high complexity. The second part still deals with the problem of interference management, but regarding with more sophisticated signal. The attention is focused on the detection of spoofing signal, which is the most complex among the jamming signal types. Due to this highly difficulty in detect and mitigate this kind of signal, spoofing threat is considered the most dangerous. In this work, a possible techniques able to detect this sophisticated signal has been proposed, observing and exploiting jointly the outputs of several operational block measurements of the GNSS receiver operating chain.

Variational and learning models for image and time series inverse problems

Inverse problems are at the core of many challenging applications. Variational and learning models provide estimated solutions of inverse problems as the outcome of specific reconstruction maps. In the variational approach, the result of the reconstruction map is the solution of a regularized minimization problem encoding information on the acquisition process and prior knowledge on the solution. In the learning approach, the reconstruction map is a parametric function whose parameters are identified by solving a minimization problem depending on a large set of data. In this thesis, we go beyond this apparent dichotomy between variational and learning models and we show they can be harmoniously merged in unified hybrid frameworks preserving their main advantages. We develop several highly efficient methods based on both these model-driven and data-driven strategies, for which we provide a detailed convergence analysis. The arising algorithms are applied to solve inverse problems involving images and time series. For each task, we show the proposed schemes improve the performances of many other existing methods in terms of both computational burden and quality of the solution. In the first part, we focus on gradient-based regularized variational models which are shown to be effective for segmentation purposes and thermal and medical image enhancement. We consider gradient sparsity-promoting regularized models for which we develop different strategies to estimate the regularization strength. Furthermore, we introduce a novel gradient-based Plug-and-Play convergent scheme considering a deep learning based denoiser trained on the gradient domain. In the second part, we address the tasks of natural image deblurring, image and video super resolution microscopy and positioning time series prediction, through deep learning based methods. We boost the performances of supervised, such as trained convolutional and recurrent networks, and unsupervised deep learning strategies, such as Deep Image Prior, by penalizing the losses with handcrafted regularization terms.

Crystal engineering strategies against antimicrobial resistance: a solid contribution to a contemporary problem

This PhD thesis sets its goal in the application of crystal engineering strategies to the design, formulation, synthesis, and characterization of innovative materials obtained by combining well established biologically active molecules and/or GRAS (generally recognized as safe) compounds with co-formers able to modulate specific properties of the molecule of interest. The solid-state association, via non-covalent interactions, of an active ingredient with another molecular component, a metal salt or a complex, may alter in a useful way the physicochemical properties of the active ingredient and/or may allow to explore new ways to enhance, in a synergistic way, the overall biological performance. More specifically this thesis will address the threat posed by the increasing antimicrobial resistance (AMR) developed by microorganisms, which call for novel therapeutic strategies. Crystal engineering provides new tools to approach this crisis in a greener and cost-effective way. This PhD work has been developed along two main research lines aiming to contribute to the search for innovative solutions to the AMR problem. Design, preparation and characterization of novel metal-based antimicrobials, whereby organic molecules with known antimicrobial properties are combined with metal atoms also known to exert antimicrobial action. Design, preparation and characterization of co-crystals obtained by combining antibacterial APIs (active pharmaceutical ingredients) with natural antimicrobials.