Artificial Intelligence-enabled Automation for Compliance Checking against GDPR

Doctoral thesis (2023)

Requirements engineering (RE) is concerned with eliciting legal requirements from applicable regulations to enable developing legally compliant software. Current software systems rely heavily on data ... [more ▼]

Requirements engineering (RE) is concerned with eliciting legal requirements from applicable regulations to enable developing legally compliant software. Current software systems rely heavily on data, some of which can be confidential, personal, or sensitive. To address the growing concerns about data protection and privacy, the general data protection regulation (GDPR) has been introduced in the European Union (EU). Organizations, whether based in the EU or not, must comply with GDPR as long as they collect or process personal data of EU residents. Breaching GDPR can be charged with large fines reaching up to up to billions of euros. Privacy policies (PPs) and data processing agreements (DPAs) are documents regulated by GDPR to ensure, among other things, secure collection and processing of personal data. Such regulated documents can be used to elicit legal requirements that are inline with the organizations’ data protection policies. As a prerequisite to elicit a complete set of legal requirements, however, these documents must be compliant with GDPR. Checking the compliance of regulated documents entirely manually is a laborious and error-prone task. As we elaborate below, this dissertation investigates utilizing artificial intelligence (AI) technologies to provide automated support for compliance checking against GDPR. • AI-enabled Automation for Compliance Checking of PPs: PPs are technical documents stating the multiple privacy-related requirements that a system should satisfy in order to help individuals make informed decisions about sharing their personal data. We devise an automated solution that leverages natural language processing (NLP) and machine learning (ML), two sub-fields of AI, for checking the compliance of PPs against the applicable provisions in GDPR. Specifically, we create a comprehensive conceptual model capturing all information types pertinent to PPs and we further define a set of compliance criteria for the automated compliance checking of PPs. • NLP-based Automation for Compliance Checking of DPAs: DPAs are legally binding agreements between different organizations involved in the collection and processing of personal data to ensure that personal data remains protected. Using NLP semantic analysis technologies, we develop an automated solution that checks at phrasal-level the compliance of DPAs against GDPR. Our solution is able to provide not only a compliance assessment, but also detailed recommendations about avoiding GDPR violations. • ML-enabled Automation for Compliance Checking of DPAs: To understand how different representations of GDPR requirements and different enabling technologies fare against one another, we develop an automated solution that utilizes a combination of conceptual modeling and ML. We further empirically compare the resulting solution with our previously proposed solution, which uses natural language to represent GDPR requirements and leverages rules alongside NLP semantic analysis for the automated support. [less ▲]

COMMUTING SATISFACTION AND SUBJECTIVE WELL-BEING: Linking life domains, workplace relocation and working from home practices

Doctoral thesis (2023)

This dissertation examines the relationship between commuting satisfaction (CS) and subjective well-being (SWB) and investigates the dynamics of commuting. For the analysis, secondary data (EU-SILC, P ... [more ▼]

This dissertation examines the relationship between commuting satisfaction (CS) and subjective well-being (SWB) and investigates the dynamics of commuting. For the analysis, secondary data (EU-SILC, P-SELL III) as well as self-collected data from an online survey about changes in workplace location and working conditions were collected. The combination of these datasets allows the exploration of three important aspects of the relationship between CS and SWB. First, the direct and indirect effects of CS on SWB are examined by considering the interplay with satisfaction with other life domains than commuting, including among others work, accommodation, time-use, leisure time, personal relationships, and health. This is an important contribution to the field of travel satisfaction because it provides an in-depth analysis of how SWB depends not only on satisfaction with a typical commute to work, but also on satisfaction with other activities that are linked to commuting. Previous studies have examined the relationship between commuting satisfaction and SWB but have largely ignored satisfaction with other life domains. This is rather surprising given that commuting depends to a large extent on decisions people make regarding other life domains such as where to live and work. This dissertation thus provides a broader conceptualization of commuting satisfaction, avoiding certain biases that otherwise might exist when interactions with satisfaction with other life domains are ignored. Second, it explores the dynamics of commuting by analyzing the impact of life events on commuting (dis)satisfaction, and the reverse. This temporal dimension of CS adds a dynamic layer to the current static interpretation of travel satisfaction by examining changes in individuals' longer-term life decisions, such as residence and/or workplace location, focusing on voluntary and involuntary relocation. Voluntary workplace relocation occurs when the employee willingly decide to change their jobs, while the latter occurs when the employee is forced to move with their employer in order to retain their jobs. This distinction in terms of workplace relocation thus provides a first empirical analysis on the dynamics of CS. Third, it allows us to examine the extent to which the relationships between CS, satisfaction with other life domains, and SWB are still applicable today, in post-pandemic times where working from home became more important than ever. This is an important contribution to the field of travel satisfaction as it provides first-hand insights into how the relationship between CS and SWB differs in post-pandemic times. The main findings from this consolidated work on travel satisfaction, particularly commuting satisfaction, are manifold. First, commuting is not a stand-alone life domain, but is connected to all other life domains, especially time-use satisfaction. Therefore, it is recommended for future studies to invest more in time-use research to understand the complexity and interplay between CS and SWB. Second, individuals who are dissatisfied with their commute do not necessarily have the financial resources and stability to change either residence or workplace to cope with dissatisfying commute patterns. These individuals who tolerate commuting dissatisfaction in their personal lives might simultaneously have a negative impact on their time-use satisfaction due to time-poverty that arises from commuting longer distances or for longer time, which obviously comes at the expense of dissatisfaction with leisure-time or personal relationships. Future research should therefore address the question of whether people make changes in their lives, for example by changing workplace location or residence, or whether they tolerate dissatisfaction with commuting, which in turn could affect their satisfaction with other life domains and SWB. This will help practitioners and policy makers in formulating the necessary transport and planning policies to accommodate these dissatisfied commuters. Fourth, people seem to be more satisfied with their commute after a voluntary workplace relocation than those who changed workplaces involuntarily. However, the question of how lasting this effect of a workplace relocation on CS is and whether CS changes over time as people become accustomed to the changed environment (treadmill effect) remains unanswered. Future research to understand the dynamics of commuting is therefore needed, using a rigorous panel design. Fifth, a workplace relocation could also lead to residential mobility. This is often noted in previous studies and somewhat addressed in this dissertation, but is not fully explored in the travel satisfaction literature. Therefore, further research is needed on the co-occurrence of life events and their impact on CS, i.e. how a workplace relocation triggers residential mobility and how lasting are its impact on CS. This can be achieved using a life-course approach to gain a better understanding of the life choices individuals make in terms of changes in their travel behavior and satisfaction, to enable better evaluation of transport and land use policies. Finally, hybrid workers (who work from home two to three days per week) seem to have higher levels of SWB compared to occasional teleworkers (who work from home less than one day per week). This implies that the well-documented relationship between CS and SWB needs to be re-examined as commuting has been limited for some people due to the outbreak of the COVID-19 pandemic as they have shifted to working from home. Future research is therefore needed to identify whether commuting actually lengthens occasional teleworkers' total workday and reduces the time they could have spent on other non-travel activities, and whether the time hybrid workers save by not commuting to work every day influences the time they spend on other non-travel activities such as household chores, childcare and sleep. Such an in-depth analysis of the interplay between CS, SWB and satisfaction with non-travel-related life domains is indeed needed to determine not only in which areas employees' well-being can be improved, but also how. On a final note, although commuting has a significant impact on individuals' SWB, it is not necessarily the most important life domain. Previous studies have shown that commuting is a stressful activity and has a direct negative impact on individual SWB; however, the results of this dissertation did not find a negative relationship between CS and SWB. In contrast to previous findings, we conclude that satisfaction with time use has the strongest total effect on SWB; regardless of how often individuals commute to work. This might suggest that individuals can maximize their utility and thus their overall SWB as long as they are free to optimize their time. As for the prospective approach of CS, we know that dissatisfaction with commute triggers changes in life event, such as (but not limited to) changing workplace or residence. However, for the majority of dissatisfied individuals who are unable to make a change, the question of how this dissatisfaction spill over onto satisfaction with non-travel-related life domains due to time poverty that results from commuting longer distances seeks further investigation. As for the dynamics, although workers who voluntary changed their workplace have higher CS than those who changes on an involuntary basis, the question of how lasting this is, and whether CS changes over time when people get accustomed to the changed environment (treadmill effect), is a topic for future research. [less ▲]

Embodied Creativity in the Era of AI: a Comparative Study involving Neural Style Transfer Technologies

Bachelor/master dissertation (2023)

Human Capital Mobility, Firm Strategy, and Innovation

Doctoral thesis (2023)

Human capital is at a pivotal position in the knowledge creation and diffusion paradigm; it is the fount of knowledge while being critical to firms’ absorption and amplification of knowledge. A firm’s ... [more ▼]

Human capital is at a pivotal position in the knowledge creation and diffusion paradigm; it is the fount of knowledge while being critical to firms’ absorption and amplification of knowledge. A firm’s sustained competitive advantage calls for organizational knowledge, which entails the integration of individual-specific knowledge embedded in employees. Yet, human beings possess attributes such as mobility and intrinsic motivation, whose implications conflict with the firm’s strategic objectives (e.g., sustaining a competitive advantage) and challenge the firm’s decision to invest in R&D. Further, given the dynamic nature of the technology market driven by constant innovation and competition, policymakers have an ongoing dialogue on how to establish sustainable innovation policies that protect the innovators’ interests without risking society’s welfare. As a result, it is of interest to managers, policymakers, and researchers to further understand the human capital position in the creation and diffusion of knowledge as well as the strategic and policy implications. This thesis contributes by exploring the reciprocal interrelations between human capital mobility, firm strategy, and innovation. Chapter 1 distinguishes voluntary from involuntary (forced) mobility and investigates the effect of forced mobility on the skilled workforce’s innovation performance from the perspectives of the source and recipient firms. We exploit the sudden fall of Nortel Networks as a natural experiment in a difference in differences event study. We construct a unique, comprehensive dataset by matching patent data from the United States Patents and Trademarks Office and online CV information from LinkedIn. We follow inventors’ careers with various employers between 1985 and 2010 and identify voluntary and involuntary (forced) movers while observing the innovation output measured by the annual number of patent applications. Our analysis shows that forced mobility distinctly affects inventors’ innovation performance when analyzed from the perspectives of the source and destination firms. Relative to voluntary mobility, the effect is insignificant, though negative, from the destination firm’s perspective. And relative to no mobility, the effect is negative, though insignificant from the destination firm’s perspective. These findings contribute to the literature on skilled workforce mobility as a driver for knowledge creation and diffusion, particularly forced mobility which has so far received very little attention. We also provide insights to policymakers and managers in the knowledge-intensive industry. Chapter 2 investigates human capital mobility behavior when unconstrained. How is the inter-firm mobility behavior of the skilled workforce altered when mobility constraints are lessened or are in favor of employees? Considering the sensitivity of knowledge sharing in the technology-intensive industry – driven by competition, constant innovation, and human capital capabilities – the mobility behavior of the skilled workforce when unconstrained is not explicit. We use changes in trade secrets protection regimes to identify mobility constraints and investigate the mobility behavior of the skilled workforce when these constraints are lessened. In a difference-in-differences setting, we find that establishing an employee-friendly trade secrets protection environment – lessening mobility constraints – reduces the likelihood of skilled workforce mobility. Further, we show that the negative effect is conveyed through the firm-specific innovation-related skills acquired by the skilled workforce. These findings align with our postulation on the increased skilled workforce’s intrinsic motivation to perform in response to the reinforced possibilities of applying the acquired technological know-how elsewhere. We illustrate the managerial dilemma of imperfect intellectual property protection and contribute to the research on intrinsic motivation as a catalyst for the observed mobility behavior and innovation output of employees working in R&D. Chapter 3 analyses firm strategy regarding R&D search and patenting while taking into account the dynamic shaping of the technology landscape by patent protection. Scholars have considerably explored the implications of patent protection and strategic patenting on innovation and R&D investment. Yet, we are still without a nuanced understanding of how firms should search the fragmented technology landscape, considering the various motives to patent and the dynamic shaping of the technology landscape by patent protection. Offensively? Defensively? Or adaptively based on payoff? We model innovation as a process of building technological bridges where new technological discoveries combine elements of new and precedent technologies. Simulation results of our model show that when patent protection shapes the technology landscape dynamically, it is advantageous for firms to search around “own” technologies globally in a complex product industry. In contrast, in a discrete product industry, it is advantageous for firms to search locally based on the payoff and globally either around “own” technologies or based on the payoff. We show that the search distance, appropriability regime, and adaptiveness of the payoff-based search influence the firms’ ultimate performance. Our study contributes to the existing research on organizational search and provides a distinct approach to modeling the structure and evolution of the technology landscape using network representation. We also provide managers and policymakers with insights regarding strategic R&D investment, patenting, and the implications on consumers and innovation welfare. [less ▲]

Transitional states in NRAS-mutated melanoma preceding drug resistance

Doctoral thesis (2023)

Patients diagnosed with advanced-stage melanoma carrying NRAS-activating mutations face a dismal prognosis, as reflected by their short progression-free survival. While targeted therapy involving MEK1/2 ... [more ▼]

Patients diagnosed with advanced-stage melanoma carrying NRAS-activating mutations face a dismal prognosis, as reflected by their short progression-free survival. While targeted therapy involving MEK1/2 and CDK4/6 co-inhibition has shown partial effectiveness in NRAS-mutant melanoma patients and has advanced to clinical trials, the molecular processes underlying acquired resistance to these drugs remain largely unknown. A triple therapeutic regimen involving inhibition of MAPK, CDK4/6, and PI3K-Akt-mTOR signalling pathways has shown promising results in impairing NRASmut melanoma progression. However, this treatment is accompanied by toxicity, highlighting the importance of identifying alternative strategies to improve efficacy and overcome the resistance of MEK and CDK4/6 inhibitors. Thus, the work performed in the framework of this doctoral thesis aimed to depict cell state transitions that propel NRASmut melanoma cells to become drug-resistant and reconstruct the cell state transcriptional trajectories associated with this phenomenon. First, we monitored cell growth and proliferation to determine the sensitivity of NRAS-mutant melanoma cell lines to MEK1/2 and CDK4/6 co-inhibition. Two main cellular responses have been observed, depending on the dynamics of the cells under drug exposure, namely fast and slow drug adaptation, and progression along the trajectory of resistance development. Next, we performed time-series single-cell RNA sequencing of distinct NRASmut melanoma cell lines over prolonged MEK/CDK4/6 co-targeting. The development of single-cell technologies has facilitated a deeper understanding of the heterogeneous transcriptional background and cell state plasticity observed in BRAFmut melanoma and to a lesser extent in NRASmut melanoma. To the best of our knowledge, this study represents the first endeavour to address the response and resistance to MEK/CDK4/6 co-inhibition at the level of single NRASmut melanoma cells. This approach enabled us to characterise cell populations that were sensitive to targeted therapy, as well as those that were intrinsically resistant or developed resistance over time. Upon early drug exposure, we detected slow-proliferating melanoma cells enriched in genes related to transmembrane transport, epithelial-mesenchymal transition (EMT), and adhesion, among others. Once melanoma cells resist inhibitory stress and resume proliferation, they transit towards a state, which is highly enriched in an interferon response gene signature. We then focused on supporting the inhibitory potential of drugs in transitional states. Activation of the transmembrane transport marker purinergic receptor P2RX7 improved the tumouricidal effects of dual MEK/CDK4/6 inhibition. Overall, this study gives a snapshot of melanoma transitional mechanisms towards resistance to targeted drugs and contributes to the establishment of novel treatment approaches for NRAS-mutant tumours. [less ▲]

LUNAR REMOTE SENSING DATA ENHANCEMENT FOR PRECISE ROBOTIC MISSION PLANNING

Doctoral thesis (2023)

In recent times, there has been a resurgence of interest in not only revisiting the Moon but also establishing a lasting and sustainable human presence there. Numer- ous agencies and private corporations ... [more ▼]

In recent times, there has been a resurgence of interest in not only revisiting the Moon but also establishing a lasting and sustainable human presence there. Numer- ous agencies and private corporations, gearing up for future lunar missions, have established key goals involve both scientific exploration and the emerging lunar economy. These objectives include carrying out scientific experiments, harvesting and utilizing lunar resources on-site, and examining potential methods of using the Moon as an efficient launchpad to go further into the Solar System. In facilitating these lunar missions, robotics emerges as the main disruptive technology. It can allow the execution of various mission-critical activities in harsh environments, ensuring mission success without jeopardizing human safety. This thesis addresses the challenge presented by the limited resolution of lu- nar data and the overwhelming amount of non-processed information provided by the different remote sensing missions. Robotic operations on the Moon require tremendous precision in the mission planning phase. For this purpose, the remote sensing data collected by various satellites must be processed and relayed to the mission planning teams in the highest possible resolution and in an easily under- standable and digested format. To address this issue, the research presented on this thesis adopts a Machine Learning (ML) approach to enhance and process lunar data gathered by different satellite sensors, providing more detailed and comprehensive insights into the lunar surface, which is crucial for future robotic mission planning. ML provides the necessary tools to efficiently handle and analyze vast volumes of data, a critical aspect in deriving meaningful results, reaching valid conclusions, and deepening our understanding of the subject under study. In this thesis, the author suggests two distinct methods for enhancing and increasing the resolution of lunar images to facilitate improved robot navigation on the lunar surface. The first method involves creating a training dataset using a digital analog environment and utilizing multiple frames of the same location for image enhancement. The second approach proposes a unique architecture that utilizes a single capture from the lunar surface for resolution upscaling, accompanied by an uncertainty estimation of the process. Lastly, a thermophysical analysis of the lunar surface is conducted, which involves processing lunar thermal data in the search for recent asteroid impacts on the lunar surface. [less ▲]

INTEGRATION AND INTERFACIAL ENGINEERING OF STRAIN SENSORS IN ADDITIVELY MANUFACTURED POLYMERIC STRUCTURES

Doctoral thesis (2023)

Additive Manufacturing (AM) (more popularly referred to as 3D printing) provides a novel solution to create multifunctional designs and parts with geometries and properties, which were impossible to ... [more ▼]

Additive Manufacturing (AM) (more popularly referred to as 3D printing) provides a novel solution to create multifunctional designs and parts with geometries and properties, which were impossible to manufacture by traditional processes. Polymer fused filament fabrication (FFF) technology as a process category of AM is currently the most popular on the market. The structure of the 3D printed neat thermoplastic polymers (TPs) and continuous fiber-reinforced thermoplastic composites (CFRTPCs) manufactured by this technology bears interfaces at different scales being the cause of non-optimal mechanical properties. Although some work has been carried out to optimize the interfacial bonding and so the structure, yet still requires further improvement. AM also opens up the possibility for embedding functional elements such as fiber Bragg grating (FBG) sensors to carry out structural health monitoring (SHM). The technology landscape of CFRTPCs created by FFF is still in an early stage of development. Further, there is also still limited work reported on interfacial engineering investigations between the polymeric matrix and the embedded FBG sensor polymer jacket for health monitoring. This dissertation studied the influence of the interfacial bonding quality on the strain transfer between the 3D printed polylactic acid (PLA) matrix and the polyimide (PI) jacketed FBG strain sensor compared to a reference strain measurement which was the digital image correlation (DIC) method. The bonding quality between the PLA matrix and the PI jacket was discussed based on the most tangible parameters which were the intrinsic adhesion and porosity. Developing bonding methodologies for improving the interfacial adhesion was therefore the first priority and were tested on PI films and PLA plates as model materials. These bonding methodologies made use of cleaning, plasma activation, roughness modification and/or the use of polydopamine (PDA) nanocoating, cyanoacrylate (CAC) glue and dichloromethane (DCM). The solvent welding by DCM and adhesive bonding with CAC and PDA as the bonding methodologies providing the highest adhesion were then implemented for the integration of a PI jacketed FBG strain sensor in PLA FFF printed structure. The main results revealed that an interfacial adhesion between the embedded FBG sensor jacket and 3D printed matrix has been necessary but not sufficient to have a good interfacial bonding since the pores’ volume in the interfacial region also contributed. Indeed, the interfacial porosity fraction which must be minimized when embedding the sensor was primarily responsible for the sensor’s strain measurement accuracy of the actual strain in the FFF printed structure. By considering the percentage of strain in the 3D printed specimen transferred to the embedded FBG strain sensor, the deposition of a PDA nanocoating onto the sensor or the addition of a CAC adhesive at the PLA-PI interface with no channel employment were identified as the best method to embed an FBG strain sensor into 3D printed PLA. The use of CAC adhesive was regarded as the most reliable method based on its lower standard deviation value for strain transfer when compared to PDA. The PDA treatment of the sensor, however, was conducted prior to printing, thus not affecting the practical aspects and with a water-based solution. Nonetheless, when selecting a methodology, the practical aspects linked to the sensor embedding in the 3D printed polymer components can also be considered based on the specific needs, and production type. To illustrate, the proper introduction of a sensor into non-linear paths in complex geometries can be facilitated by a channel. That being the case, the combination of a channel and a CAC adhesive could be considered to best suit. The original findings presented in this thesis explain the link between the matrix-sensor effective strain transfer and interfacial properties and demonstrate that the reliability of an embedded FBG sensor’s measurement depends on its good bonding with the surrounding matrix. The results presented shed light on the understanding of the importance of interfacial engineering in FFF built structures with FBG sensors, which is a fundamental step in this technology towards the development of more efficient smart polymer structures. [less ▲]

HEAT CURRENTS, WORK STATISTICS AND FLUCTUATIONS IN QUANTUM SYSTEMS

Doctoral thesis (2023)

Resonant Raman scattering and other new coupling phenomena in ferroelastic BiVO4

Doctoral thesis (2023)

With the introduction of ferroic oxides into the research field of photovoltaic and photocatalytic applications, an increasing interest in understanding the coupling mechanisms of light with the ... [more ▼]

With the introduction of ferroic oxides into the research field of photovoltaic and photocatalytic applications, an increasing interest in understanding the coupling mechanisms of light with the electronic bands in this class of materials has been developed over the last years. In many of these examples, reliable band structure calculations are not yet available mostly because of the complexity in the crystal structure which demands for many experimental evidences to verify even basic characteristics of the structure. Also classical experimental techniques based on the optical absorption seem to reach their limits here resulting often in controversial publications on the nature of the band gap. The goal of this work is to exploit resonant Raman scattering to reveal information on the electronic band structure and thereby create a better understanding of the light-matter coupling mechanisms in ferroic oxides. As a model material, we choose to work on ferroelastic bismuth vanadate, which is known for its second-order phase transition from the high-symmetry tetragonal to the low-symmetry monoclinic phase. On the way to understand resonant Raman scattering in bismuth vanadate, i.e. the coupling of lattice vibrations and electronic transitions, we extensively studied the phonons and optical properties and discovered new coupling phenomena of various kinds. We found a phonon-phonon coupling in the lattice vibrations of the vanadium oxygen tetrahedron that are strongly temperature and polarization dependent. The coupling strength is of opposite sign depending on the polarization conditions and diminishes at the structural phase transition at which one of the phonon modes changes its symmetry. Additionally, the transmission data evidence the coupling between the spontaneous shear strain and the electronic structure that primarily defines the strong temperature dependence of the optical absorption in monoclinic bismuth vanadate. Further, we report on resonant Raman scattering effects that allow us to study the coupling between phonons and electrons. We succeeded to measure multiple resonant Raman bands corresponding to not only the band gap at 2.4 eV but also to a polaronic defect level at 2.0 eV. The coupling strength to the electronic states in the conduction band and in the defect level varies between the Raman modes corresponding to different lattice vibrations. By varying the light polarization in the experiment, we could access another band-to-band transition that lies 120 meV above the band gap energy. This energy difference matches the optical anisotropy quantified by transmission measurements. With this work we demonstrate that resonant Raman spectroscopy is a very powerful tool to study not only structural but also electronic properties in ferroic oxides. It has the potential to go deeper in the analysis on the electronic band structure than classical techniques such as UV/Vis spectroscopy or ellipsometry. [less ▲]

(DE-) SCAFFOLDING AND THE EVOLUTION OF AUTONOMY IN THE TEACHING AND LEARNING PROCESS - AN INTERGENERATIONAL CASE STUDY

Doctoral thesis (2023)

Psychology has produced many theories that do not concern themselves (enough) with the fact that human beings are constantly moving. This thesis takes psychological theories and empirical data gathering ... [more ▼]

Psychology has produced many theories that do not concern themselves (enough) with the fact that human beings are constantly moving. This thesis takes psychological theories and empirical data gathering on to the slopes of the Austrian Alps. Alpine downhill skiing is viewed from different psychological points – teaching/learning with the notion of scaffolding, moving in a special environment, family dynamics using Dialogical Self theory, resilience. Empirical data was collected in a longitudinal single case study over the span of four years. A combination of interviews, fill-in-the-blanks texts, and videos was used to grasp the complex reality of going down the slope. Action cameras are introduced as a practical way to gather unique data from the participants’ point of view without the disturbance of the researcher’s presence. Double Direction Theme Completion (DDTC) method was used for the fill-in-the-blanks to provoke and show ambivalence parents experience when teaching their children how to ski. The interviews were held using the method of the narrative interview to obtain long passages of storytelling. This cumulative thesis consists of three peer-reviewed papers and one book chapter. This project showed a way to bring more movement into psychological theories and how to gather empirical data in moving research settings. Outdoor settings have proven to provide interesting new data to analyse using already existing theories like scaffolding, dialogical self theory (DST), or resilience. The use of DST was expanded from mainly therapy settings and classrooms to an action setting. The unique conditions of demanding environments like the slope challenge I-Position to change which can be examined using the DDTC method. The notion of descaffolding is introduced in this thesis. Descaffolding describes the process of withdrawing support in a learning setting and setting the stage for autonomous mastery. The descaffolding process is described in detail and further applications are discussed. [less ▲]

Aiming towards more equal opportunities? Analyzing the implementation of the plurilingual education policy in the non-formal early childhood education and care sector in Luxembourg

Doctoral thesis (2023)

The plurilingual education policy in Luxembourg stands out as one of the pioneering multilingual policies in early childhood education that goes beyond a bilingual approach implemented in other countries ... [more ▼]

The plurilingual education policy in Luxembourg stands out as one of the pioneering multilingual policies in early childhood education that goes beyond a bilingual approach implemented in other countries. On this basis, given that investment in early childhood education and care aims to tackle social and origin-related inequalities and that educational inequalities in Luxembourg are frequently attributed to multilingualism, promoting children’s language development is at the centre of early childhood. This interdisciplinary research project fills a significant research gap in the social science fields of policy implementation and early childhood by focusing on the aspect of implementation in relation to educational inequalities. This study answers the following research questions: (I) how is the plurilingual education policy implemented in the Luxembourgish non-formal early childhood education and care sector, and (II) what is the salience of the aspect of reducing educational inequalities. The theoretical lenses employed in this study encompass concepts relevant to both aspects of the study. On the one hand, (I) the coupling theory, theory of planned behavior, and theory of habitus and field, and, on the other hand, (II) the reproduction of educational inequality, linguistic capital theory, and theory of social and ethnic origin. By using a mixed methods design, examining both policy and practice levels provides a holistic perspective on the plurilingual education policy implementation in Luxembourg. Methods include a policy document analysis of the plurilingual education program, expert interviews with policy-level stakeholders, and a cross-sectional survey with early childhood practitioners in the non-formal education sector. The main findings of this research shed light on sectorial differences at policy and practice levels, leading to ambiguities in policy goals and implementation measures, as well as challenges encountered during policy implementation. The results indicate a lack of an epistemological approach in the policy, regarding the linguistic and organizational diversity within the Luxembourgish mixed economy of early childhood education and care. Concerns arise regarding the goal of promoting more equal opportunities, as the policy may – according to the findings – still provide greater benefits to Luxembourgish-speaking children rather than to other ethnic groups. The research thus highlights the importance of developing multilingual policies that consider the diverse linguistic and organizational contexts, in order to ensure a more successful policy in practice implementation as well as more equal educational opportunities for all children. [less ▲]

Formal Verification of Verifiability in E-Voting Protocols

Doctoral thesis (2023)

Election verifiability is one of the main security properties of e-voting protocols, referring to the ability of independent entities, such as voters or election observers, to validate the outcome of the ... [more ▼]

Election verifiability is one of the main security properties of e-voting protocols, referring to the ability of independent entities, such as voters or election observers, to validate the outcome of the voting process. It can be ensured by means of formal verification that applies mathematical logic to verify the considered protocols under well-defined assumptions, specifications, and corruption scenarios. Automated tools allow an efficient and accurate way to perform formal verification, enabling comprehensive analysis of all execution scenarios and eliminating the human errors in the manual verification. The existing formal verification frameworks that are suitable for automation are not general enough to cover a broad class of e-voting protocols. They do not cover revoting and cannot be tuned to weaker or stronger levels of security that may be achievable in practice. We therefore propose a general formal framework that allows automated verification of verifiability in e-voting protocols. Our framework is easily applicable to many protocols and corruption scenarios. It also allows refined specifications of election procedures, for example accounting for revote policies. We apply our framework to the analysis of several real-world case studies, where we capture both known and new attacks, and provide new security guarantees. First, we consider Helios, a prominent web-based e-voting protocol, which aims to provide end-to-end verifiability. It is however vulnerable to ballot stuffing when the voting server is corrupt. Second, we consider Belenios, which builds upon Helios and aims to achieve stronger verifiability, preventing ballot stuffing by splitting the trust between a registrar and the server. Both of these systems have been used in many real-world elections. Our third case study is Selene, which aims to simplify the individual verification procedure for voters, providing them with trackers for verifying their votes in the clear at the end of election. Finally, we consider the Estonian e-voting protocol, that has been deployed for national elections since 2005. The protocol has continuously evolved to offer better verifiability guarantees but has no formal analysis. We apply our framework to realistic models of all these protocols, deriving the first automated formal analysis in each case. As a result, we find several new attacks, improve the corresponding protocols to address their weakness, and prove that verifiability holds for the new versions. [less ▲]

What Matters in Model Training to Transfer Adversarial Examples

Doctoral thesis (2023)

Despite state-of-the-art performance on natural data, Deep Neural Networks (DNNs) are highly vulnerable to adversarial examples, i.e., imperceptible, carefully crafted perturbations of inputs applied at ... [more ▼]

Despite state-of-the-art performance on natural data, Deep Neural Networks (DNNs) are highly vulnerable to adversarial examples, i.e., imperceptible, carefully crafted perturbations of inputs applied at test time. Adversarial examples can transfer: an adversarial example against one model is likely to be adversarial against another independently trained model. This dissertation investigates the characteristics of the surrogate weight space that lead to the transferability of adversarial examples. Our research covers three complementary aspects of the weight space exploration: the multimodal exploration to obtain multiple models from different vicinities, the local exploration to obtain multiple models in the same vicinity, and the point selection to obtain a single transferable representation. First, from a probabilistic perspective, we argue that transferability is fundamentally related to uncertainty. The unknown weights of the target DNN can be treated as random variables. Under a specified threat model, deep ensemble can produce a surrogate by sampling from the distribution of the target model. Unfortunately, deep ensembles are computationally expensive. We propose an efficient alternative by approximately sampling surrogate models from the posterior distribution using cSGLD, a state-of-the-art Bayesian deep learning technique. Our extensive experiments show that our approach improves and complements four attacks, three transferability techniques, and five more training methods significantly on ImageNet, CIFAR-10, and MNIST (up to 83.2 percentage points), while reducing training computations from 11.6 to 2.4 exaflops compared to deep ensemble on ImageNet. Second, we propose transferability from Large Geometric Vicinity (LGV), a new technique based on the local exploration of the weight space. LGV starts from a pretrained model and collects multiple weights in a few additional training epochs with a constant and high learning rate. LGV exploits two geometric properties that we relate to transferability. First, we show that LGV explores a flatter region of the weight space and generates flatter adversarial examples in the input space. We present the surrogate-target misalignment hypothesis to explain why flatness could increase transferability. Second, we show that the LGV weights span a dense weight subspace whose geometry is intrinsically connected to transferability. Through extensive experiments, we show that LGV alone outperforms all (combinations of) four established transferability techniques by 1.8 to 59.9 percentage points. Third, we investigate how to train a transferable representation, that is, a single model for transferability. First, we refute a common hypothesis from previous research to explain why early stopping improves transferability. We then establish links between transferability and the exploration dynamics of the weight space, in which early stopping has an inherent effect. More precisely, we observe that transferability peaks when the learning rate decays, which is also the time at which the sharpness of the loss significantly drops. This leads us to propose RFN, a new approach to transferability that minimises the sharpness of the loss during training. We show that by searching for large flat neighbourhoods, RFN always improves over early stopping (by up to 47 points of success rate) and is competitive to (if not better than) strong state-of-the-art baselines. Overall, our three complementary techniques provide an extensive and practical method to obtain highly transferable adversarial examples from the multimodal and local exploration of flatter vicinities in the weight space. Our probabilistic and geometric approaches demonstrate that the way to train the surrogate model has been overlooked, although both the training noise and the flatness of the loss landscape are important elements of transfer-based attacks. [less ▲]

Development of the bioinformatics pipeline DREMflow for the identification of cell-type and time point specific transcriptional regulators

Doctoral thesis (2023)

A detailed understanding of the mechanism that drive cell differentiation of stem cells into a desired cell type provides opportunities to study diseases and disease progression in patient derived cells ... [more ▼]

A detailed understanding of the mechanism that drive cell differentiation of stem cells into a desired cell type provides opportunities to study diseases and disease progression in patient derived cells and enable the development of new therapy approaches. The main challenge in this directed differentiation is the identification of the essential transcriptional regulators involved that are specific to a cell type or lineage and the inference of the underlying gene regulatory network. Transcription factor activity during cell differentiation can be measured through gene expression and chromatin accessibility, ideally jointly over time. Integrated time course regulatory analysis yields more detailed gene regulatory networks than expression data alone. Due to the large number of parameters and tools employed in such analysis, computational workflows help to manage the inherent complexity of such analyses. This thesis describes Dynamics Regulatory Events Miner Snakemake workflow (DREMflow) which combines temporally-resolved RNA-seq and ATAC-seq data to identify cell type and time point specific gene regulatory networks. DREMflow builds on the Differentially Regulatory Events Miner (DREM), the workflow management system Snakemake and the package manager Mamba. It includes the processing starting from sequencing reads, quality control reports and parameters as well as additional downstream analyses for the inference of key transcription factors during differentiation. DREMflow is applied to multiple data sets obtained during the differentiation of midbrain dopaminergic neurons as well as blood cells and compared to TimeReg, a pipeline with similar aims. The expansion to accommodate for single-cell data is explored. Results from other studies were reproduced and extended, identifying additional key transcriptional regulators. LBX1 was found as key regulator in differentiation of midbrain dopaminergic neurons while exploring different settings of the pipeline. Members of the AP-1 family of transcription factors were identified in all blood cell differentiation data sets. The comparison to TimeReg resulted in DREMflow being more sensitive in the identification of known transcriptional regulators in macrophages. Computationally, DREMflow outperforms TimeReg as well. DREMflow enables users to perform time-resolved multi-omics analysis reproducibly with minimal setup and configuration. [less ▲]

Spatial adaptive settlement systems in archaeology. Modelling long-term settlement formation from spatial micro interactions

Doctoral thesis (2023)

Despite research history spanning more than a century, settlement patterns still hold a promise to contribute to the theories of large-scale processes in human history. Mostly they have been presented as ... [more ▼]

Despite research history spanning more than a century, settlement patterns still hold a promise to contribute to the theories of large-scale processes in human history. Mostly they have been presented as passive imprints of past human activities and spatial interactions they shape have not been studied as the driving force of historical processes. While archaeological knowledge has been used to construct geographical theories of evolution of settlement there still exist gaps in this knowledge. Currently no theoretical framework has been adopted to explore them as spatial systems emerging from micro-choices of small population units. The goal of this thesis is to propose a conceptual model of adaptive settlement systems based on complex adaptive systems framework. The model frames settlement system formation processes as an adaptive system containing spatial features, information flows, decision making population units (agents) and forming cross scale feedback loops between location choices of individuals and space modified by their aggregated choices. The goal of the model is to find new ways of interpretation of archaeological locational data as well as closer theoretical integration of micro-level choices and meso-level settlement structures. The thesis is divided into five chapters, the first chapter is dedicated to conceptualisation of the general model based on existing literature and shows that settlement systems are inherently complex adaptive systems and therefore require tools of complexity science for causal explanations. The following chapters explore both empirical and theoretical simulated settlement patterns based dedicated to studying selected information flows and feedbacks in the context of the whole system. Second and third chapters explore the case study of the Stone Age settlement in Estonia comparing residential location choice principles of different periods. In chapter 2 the relation between environmental conditions and residential choice is explored statistically. The results confirm that the relation is significant but varies between different archaeological phenomena. In the third chapter hunter-fisher-gatherer and early agrarian Corded Ware settlement systems were compared spatially using inductive models. The results indicated a large difference in their perception of landscape regarding suitability for habitation. It led to conclusions that early agrarian land use significantly extended land use potential and provided a competitive spatial benefit. In addition to spatial differences, model performance was compared and the difference was discussed in the context of proposed adaptive settlement system model. Last two chapters present theoretical agent-based simulation experiments intended to study effects discussed in relation to environmental model performance and environmental determinism in general. In the fourth chapter the central place foragingmodel was embedded in the proposed model and resource depletion, as an environmental modification mechanism, was explored. The study excluded the possibility that mobility itself would lead to modelling effects discussed in the previous chapter. The purpose of the last chapter is the disentanglement of the complex relations between social versus human-environment interactions. The study exposed non-linear spatial effects expected population density can have on the system and the general robustness of environmental inductive models in archaeology to randomness and social effect. The model indicates that social interactions between individuals lead to formation of a group agency which is determined by the environment even if individual cognitions consider the environment insignificant. It also indicates that spatial configuration of the environment has a certain influence towards population clustering therefore providing a potential pathway to population aggregation. Those empirical and theoretical results showed the new insights provided by the complex adaptive systems framework. Some of the results, including the explanation of empirical results, required the conceptual model to provide a framework of interpretation. [less ▲]

Correspondent banking, SWIFT, and the geographies of financial infrastructure: Technological and organizational change in cross-border payments

Doctoral thesis (2023)

This thesis examines the impacts of technical and organizational change on the geographies of finance via infrastructure for cross-border payments, employing a qualitative methodology of semi-structured ... [more ▼]

This thesis examines the impacts of technical and organizational change on the geographies of finance via infrastructure for cross-border payments, employing a qualitative methodology of semi-structured expert interviews. The study finds that SWIFT’s messaging system together with the correspondent banking system, a decentralized global network of bilateral contracts between banks, remain a geographically and historically foundational sociotechnical infrastructure connecting IFCs. To stave off fintech challengers and preserve banks’ incumbency, SWIFT’s system is platformizing with the aim of changing banks’ business models from fee-extraction towards economic use of transaction data. Collaborative action in bringing about change across a global network is a key finance industry agency for maintaining its collective dominance. SWIFT’s cooperative organizational form is a significant locus for this agency, engendering trust as a relational aspect of power to resolve tensions among actors and processes across scales. Specialized infrastructure is instrumental in how the geographies of finance are (re)shaped. [less ▲]

Analysis of Smartcard-based Payment Protocols in the Applied Pi-calculus using Quasi-Open Bisimilarity

Doctoral thesis (2023)

Cryptographic protocols are instructions explaining how the communication be- tween agents should be done. Critical infrastructure sectors, such as communication networks, financial services, information ... [more ▼]

Cryptographic protocols are instructions explaining how the communication be- tween agents should be done. Critical infrastructure sectors, such as communication networks, financial services, information technology, transportation, etc., use security protocols at their very core to establish the information exchange between the components of the system. Symbolic verification is a discipline that investigates whether a given protocol satisfies the initial requirements and delivers exactly what it intends to deliver. An immediate goal of symbolic verification is to improve the reliability of existing systems – if a protocol is vulnerable, actions must be taken asap before a malicious attacker exploits it; a far-reaching goal is to improve the system design practices – when creating a new protocol, it must be proven correct before the implementation. Properties of cryptographic protocols roughly fall into two categories. Either reachability-based, i.e. that a system can or cannot reach a state satisfying some condition, or equivalence-based, i.e. that a system is indistinguishable from its idealised version, where the desired property trivially holds. Security properties are often formulated as a reachability problem and privacy properties as an equivalence problem. While the study of security properties is relatively settled, and powerful tools like Tamarin and ProVerif, where it is possible to check reachability queries, exist, the study of privacy properties expressed as equivalence only starts gaining momentum. Tools like DeepSec, Akiss, and, again, ProVerif offer only limited support when it comes to indistinguishability. This is partially due to the question of “What is an attacker capable of?” is not answered definitively in the second case. The widely-accepted default attacker, when it comes to security, is the so-called Dolev-Yao attacker, which has full control of the communication network; however, there is no default attacker who attempts to break the privacy of a protocol. The capabilities of such an attacker are reflected in the equivalence relation used to define a privacy property; hence the choice of such relation is crucial. This dissertation justifies a particular equivalence relation called quasi-open bisimilarity which satisfies several natural requirements. It has sound and complete modal logic characterisation, meaning that any attack on privacy has a practical interpretation; it enables compositional reasoning, meaning that if a privacy property of a system automatically extends to a bigger system having the initial one as a component, and, it captures the capability of an attacker to make decisions dynamically during the execution of the protocol. We not only explain the notion of quasi-open bisimilarity, but we also employ it to study real-world protocols. The first protocol, UBDH, is an authenticated key agreement suitable for card payments, and the second protocol, UTX, is a smartcard-based payment protocol. Using quasi-open bisimilarity, we define the target privacy property of unlinkability, namely that it is impossible to link protocol sessions made with the same card and prove that it holds for UBDH and UTX. The proofs that UBDH and UTX satisfy their privacy requirements to our knowledge are the first ones that demonstrate that a privacy property of a security protocol, defined as bisimilarity equivalence, is satisfied for an unbounded number of protocol sessions. Moreover, these proofs illustrate the methodology that could be employed to study the privacy of other protocols. [less ▲]

A BIOMECHANICAL STUDY OF THE PELVIS WITH AND WITHOUT FRACTURES AND IMPLANTS: COMBINING COMPUTATIONAL DESIGN AND EXPERIMENTAL TESTING FOR TYPICAL DAILY MOVEMENTS

Doctoral thesis (2023)

This study fulfills the need for a dedicated pelvis testing setup that is widely accepted with physiological relevance. There is limited comparative biomechanical data available for the Supraacetabular ... [more ▼]

This study fulfills the need for a dedicated pelvis testing setup that is widely accepted with physiological relevance. There is limited comparative biomechanical data available for the Supraacetabular External Fixator (SEF) and the Subcutaneous Iliopubic Plate (SIP) used in the treatment of anterior fragility fractures of the pelvis (FFP). Most experimental studies so far have relied on simplified loading and boundary conditions. There has been a growing interest in personalizing motion analysis to develop customized implants that can optimize implant performance and accelerate fractured bone recovery for individual patients. Therefore, the main objective of this study is to develop a biomechanical test bench that can emulate the physiological gait loading of the pelvis, experimentally evaluate the stabilizing effect of the SEF and the SIP in the treatment of FFP, expand the test stand's capability to emulate other common daily movements, investigate the impact of customized musculoskeletal (MS) models, and assess the potential benefits of using personalized 3D metallic printed subcutaneous plates for the treatment of FFP type Ia fractures. The study uses the Computational Experiment Design procedure to design a biomechanical test stand that realistically emulates the pelvis' physiological gait loading. The test stand is designed to iteratively reduce all muscles and joints' contact forces of the pelvis to only four force actuators while still producing a similar stress distribution in the pelvis. The study conducts repeatability and reproducibility tests to ensure the test stand's capabilities. Next, the FFP type Ia is created on a synthetic pelvis for biomechanical testing under gait loading. The osteotomy on the right pelvic ring is then stabilized with the SEF or the nonlocking/locking SIP, and the stability provided by both implants is assessed numerically and experimentally under physiological loading. Motion analysis is conducted to calculate joint and muscle force envelopes for the common daily movement of interest. Stress, strain and displacement of the pelvis under these loads are assessed numerically and then implemented in the biomechanical test stand to emulate each movement following the computational experiment design concept. A metallic 3D-printed SIP is developed to match the anatomical landmarks of the insertion points on the pelvis used in the experiments. This 3D printed plate is assessed numerically and experimentally under physiological load to evaluate its performance compared to conventional plates. Personalization of the MS model is conducted for the pelvis by matching the anatomical landmarks of the pelvis in the generic MS model and the model of the pelvis used in the actual experiments. The developed test stand and the concept of computational experiment design behind it provide guidelines on how to design biomechanical testing equipment with physiological relevance. The boundary conditions and the nature of loading adopted in this study are more realistic regarding physiological relevance compared to the state-of-the-art. The numerically developed biomechanical testing setup of the pelvis in this study is a significant step forward in developing a physiologically relevant pelvis testing setup. [less ▲]

Enabling Resilient and Efficient Communication for the XRP Ledger and Interledger

Doctoral thesis (2023)

The blockchain technology is relatively new and still evolving. Its development was fostered by an enthusiastic community of developers, which sometimes forgot about the lessons from the past related to ... [more ▼]

The blockchain technology is relatively new and still evolving. Its development was fostered by an enthusiastic community of developers, which sometimes forgot about the lessons from the past related to security, resilience and efficiency of communication which can impact network scalability, service quality and even service availability. These challenges can be addressed at network level but also at operating system level. At network level, the protocols and the architecture used play a major role, and overlays have interesting advantages like custom protocols and the possibility of arbitrary deployments. This thesis shows how overlay networks can be designed and deployed to benefit the security and performance in communication for consensus-validation based blockchains and blockchain inter-operativity, taking as concrete cases the XRP ledger and respectively the Interledger protocol. XRP Ledger is a consensus-validation based blockchain focused on payments which currently uses a flooding mechanism for peer to peer communication, with a negative impact on scalability. One of the proposed overlays is based on Named Data Networking, an Internet architecture using for propagation the data name instead of data location. The second proposed overlay is based on Spines, a solution offering improved latency on lossy paths, intrusion tolerance and resilience to routing attacks. The system component was also interesting to study, and the contribution of this thesis centers around methodologies to evaluate the system performance of a node and increase the security from the system level. The value added by the presented work can be synthesized as follows: i) investigate and propose a Named Data Networking-based overlay solution to improve the efficiency of intra-blockchain communication at network level, taking as a working case the XRP Ledger; ii) investigate and propose an overlay solution based on Spines, which improves the security and resilience of inter-blockchain communication at network level, taking as a working case the Interledger protocol; iii) investigate and propose a host-level solution for non-intrusive instrumentation and monitoring which helps improve the performance and security of inter-blockchain communication at the system level of machines running Distributed Ledger infrastructure applications treated as black-boxes, with Interledger Connectors as a concrete case. [less ▲]

Unravelling the early pathological mechanisms in Parkinson's Disease: Insights from alpha-synuclein dependent and independent models

Doctoral thesis (2023)

Affecting over 10 million people worldwide, Parkinson’s disease is the second most common neurodegenerative disorder. With only 10% of cases having a known genetic cause, PD aetiology largely remains an ... [more ▼]

Affecting over 10 million people worldwide, Parkinson’s disease is the second most common neurodegenerative disorder. With only 10% of cases having a known genetic cause, PD aetiology largely remains an enigma. Endogenous factors such as genetic predisposition, and exogenous factors such as exposure to toxins and lifestyle choices, interplay in the initiation and acceleration of the disease. Despite some common hallmarks such as nigrostriatal degeneration and Lewy bodies pathology, PD clinical picture largely varies across patients. Non-motor symptoms are common and thought to emerge up to 20 years prior to diagnosis, and they range from gastro-intestinal dysfunction to sleep disturbances to hallucinations. 90% of PD patients present at least one neuropsychiatric symptom, and about 30% of total patients develop dementia. Interventions aimed to prevent or slowdown disease progression require a better understanding of the early molecular events which foster neuronal dysfunction and death. Longitudinal studies which allow the investigation of early pathological stages are challenging to achieve on patients-based study only, thus largely rely on the use of animal models. Specifically, rodents have very similar anatomy, physiology, and genetics to humans, and a good set of genetic/molecular tools are available to generate pathological models. In the present thesis, we ventured into the investigation of alpha-synuclein dependent and independent models of PD, to unravel the early molecular events driving PD pathogenesis. Firstly, we investigated a genetic mouse model overexpressing the human, E46K mutated alpha-synuclein gene. We characterised neurodegeneration in the nigrostriatal pathway and motor deficits, detecting characteristics of an early-PD phenotype. Aiming to understand the molecular events driving neurodegeneration, we profiled the ventral midbrain transcriptome at different ages, uncovering that transcriptional changes are an early response to the alpha-synuclein challenge. Being the E46K mutation associated with dementia, we also profiled the hippocampus to investigate early transcriptional events linked with cognitive dysfunction in PD. We revealed that hippocampal dysfunction is mostly driven by the ageing process, operating over the interplay of genetic and gender predisposition. Secondly, we profiled transcriptomic changes in the midbrain of the alpha-synuclein independent, Park7-/- (DJ-1 KO) mouse model. Once again, we uncovered the interplay of sex and age in determining the susceptibility to the disease challenge, with males being more affected than females. Specifically, the response to DJ-1 loss of function appeared to be largely sex-specific, and to be mediated by the oestrogen pathway and the DJ-1/Nrf2/CYP1B1 axis. Even if sex-dimorphism has not been directly investigated in human Park7 PD cases due to their paucity, it has been reported in sporadic PD for several populations. Thus, our findings might significantly contribute to uncovering the reasons behind gender differences in PD. Thirdly, we investigated a moderate overexpressor of wild-type alpha-synuclein (Thy1-Syn14), aiming to reach a compromise between genetic and idiopathic PD modelling. To understand how endogenous and exogenous factors interplay in disease onset and progression, we exposed transgenic mice to the amyloidogenic protein Curli and to a fibre deprived diet. We uncovered that microbiome insults and diet act in combination to promote disease progression, and we provided supporting evidence to the concept of a gut-brain axis in PD. Our results underline the relevance of lifestyle adjustments in the management of PD patients. Finally, we investigated how different alpha-synuclein moieties and glutamate exposure might contribute to neurodegeneration. Even if these studies were left incomplete, we gained some preliminary indications which can represent a starting point for future research. We observed that both oligomers and fibrils are toxic forms of alpha-synuclein, and that a lack of standardisation in recombinant moieties production is a current issue that may halt reproducibility in alpha-synuclein research. We also reported a higher susceptibility of DJ-1 knock-down cells to glutamate excitotoxicity, potentially underlying an additional mechanism through which DJ-1 loss of function is responsible for PD development. [less ▲]

STUDYING THE IMPACT OF A53T α-SYNUCLEIN ON ASTROCYTIC FUNCTIONS AND ACTIVATION IN HUMAN IPSC-DERIVED CULTURES

Doctoral thesis (2023)

With its high prevalence among the elderly, the movement disorder Parkinson’s disease (PD) poses a major challenge for our society. Unfortunately, despite continuous efforts from the research community ... [more ▼]

With its high prevalence among the elderly, the movement disorder Parkinson’s disease (PD) poses a major challenge for our society. Unfortunately, despite continuous efforts from the research community, we still lack the disease-modifying treatments for this condition. Therefore, it is of great importance to develop suitable models, which can be employed to better understand the molecular mechanisms underlying PD. In this context, iPSC technology offers a possibility to study the disease pathogenesis using patient-derived brain cells. In recent years, astrocytes have come into the spotlight as potential major contributors to PD development. Yet, there is a limited number of studies utilizing iPSC-derived models to examine PD-linked mutations at endogenous levels. This thesis aims to address the described gap by studying the effect of the A53T α-synuclein on the physiology of human iPSC-derived astrocytes. To identify a suitable model, we first compared two published protocols for the generation of iPSC astrocytes, referred to as Oksanen and Palm method, respectively. A transcriptomic analysis revealed higher maturation characteristics for Oksanen astrocytes. Furthermore, these astrocytes showed a higher similarity to their human postmortem counterparts. Applying the Oksanen protocol, we generated astrocytes derived from a healthy individual and a patient carrying the G209A mutation in SNCA, corresponding to p.A53T substitution in α-synuclein. The utilization of single-cell RNA sequencing allowed us to identify perturbed molecular mechanisms exclusively in pure astrocytic populations. We could demonstrate that astrocytes have a decreased capacity to differentiate. Furthermore, we observed a distinct response of the two cell lines to triggers of activation. Interestingly, activated patient astrocytes also showed changes in pathways related to mitochondrial homeostasis. Taken together, we show that A53T α-synuclein has a profound effect on the function of iPSC-derived astrocytes. In particular, we could demonstrate that patient astrocytes differ from healthy control cells in their activation status and with respect to mitochondrial biology. Further investigation will be required to elucidate the impact of the identified perturbations on the astrocyte-neuron interplay in the context of PD. [less ▲]

Supporting Safety Analysis of Deep Neural Networks with Automated Debugging and Repair

Doctoral thesis (2023)

EXAMINATION OF ROLE AND FUNCTION OF DJ-1 IN PARKINSON’S DISEASE AND GLIOBLASTOMA

Doctoral thesis (2023)

AGE- AND SEX-SPECIFIC TRANSCRIPTOME CHANGES IN THE MIDBRAIN OF PARK7-DEPLETED MICE

Doctoral thesis (2023)

To date, various functions have been reported for DJ-1 protein, encoded by PARK7 gene, mostly associated to maintenance of a balance between the reactive oxygen species (ROS) production and the ... [more ▼]

To date, various functions have been reported for DJ-1 protein, encoded by PARK7 gene, mostly associated to maintenance of a balance between the reactive oxygen species (ROS) production and the antioxidant response. Indeed, DJ-1 has been related with several oxidative-stress associated diseases, either by its overexpression or by its absence. Loss-of-function mutations in PARK7 can lead to an early onset PD. PD is the second most common neurodegenerative disease that usually affects the population above the age of 65. PD is characterized by its motor symptoms, although prodromal non-motor symptoms can appear up to 20 years earlier and can have a major impact on quality of life for the patients. One of the most characteristic neuropathological hallmarks of PD is the progressive dopaminergic neuronal loss in the substantia nigra pars compacta (SNpc) of the midbrain, leading to striatal dopamine deficiency. One of the main drivers of the disease is oxidative stress caused by mitochondrial dysfunction. Sex-differences in the incidence, prevalence and severity of the disease are observed in PD, with males affected more than females. In mice, Park7 deletion leads to dopaminergic deficits during aging, and increased sensitivity to oxidative stress. However, the severity of the reported phenotypes varies, and the findings are often not separated by sex. In the present study, gene expression signatures of in vivo midbrain sections from male and female Park7 knock-out mice were investigated at different ages to understand the early, prodromal molecular changes upon loss of DJ-1. Interestingly, while at 3 months the transcriptomes of both male and female mice were unchanged compared to their wild type littermates, an extensive deregulation was observed specifically in 8-month-old males. The affected genes were enriched for processes such as focal adhesion, extracellular matrix (ECM) interaction, and epithelial-to-mesenchymal transition (EMT), while the most enriched transcription factor at the deregulated genes was nuclear factor erythroid 2-related factor 2 (NRF2). Among others, the EMT marker gene Cdh1 as well as antioxidant response genes were altered specifically in the midbrain, but not in the cortex, of male DJ-1 deficient mice. Moreover, many of the misregulated genes are known target genes of estradiol (E2) and all-trans-retinoic acid (ATRA) signaling and show sex-specific expression in wild type mice. In line with this, downregulation of the expression of Cyp1b1, encoding an enzyme involved in the metabolism of both E2 and ATRA was also observed only in the midbrain of male DJ-1 deficient mice. Depletion of DJ-1 or NRF2 in primary male astrocytes recapitulated many of the in vivo changes, including downregulation of Cyp1b1. Interestingly, knock-down of Cyp1b1 led to gene expression changes in focal adhesion and EMT in cultured male astrocytes. Moreover, iPSC-derived astrocytes from PD patient with loss-of-function PARK7 mutation showed changes in genes associated with EMT pathway and NRF2 signaling. Taken together, our data indicate that loss of Park7 leads to sex-specific gene expression changes specifically in males through astrocytic alterations in NRF2-CYP1B1 axis. In addition, since an extensive deregulation occurs in the midbrain of 8-month-old males, the single nuclei chromatin accessibility profile at this age and sex was also investigated, with the aim of identifying the upstream regulatory events that lead to the observed transcriptomic changes and the implication of each cell type on those changes. Despite the low number of recovered nuclei, the major brain cell types were successfully identified. Similar representation in terms of proportion of the total nuclei was observed between the genotypes for all identified cell type populations except for astrocytes, that showed lower numbers in Park7-/- mice in comparison to wild type mice. Moreover, the biggest differences in chromatin accessibility were observed in the astrocyte population that also showed the strongest overlap with the transcriptomic changes. Enrichment analysis performed over the genes showing both epigenomic and transcriptomic changes in astrocytes were consistent with pathways identified in the analysis of the entire midbrain sections. Further indicating the relevance of these changes and their association with astrocytes in the mouse midbrain with Park7 depletion at both chromatin and mRNA level. These findings provide new information about Park7-/- PD mouse model, showing specific changes during aging and suggesting higher sensitivity of males to loss of DJ-1, which might help to better understand variation in the reported phenotypes of Park7-/- mice. These results also point to astrocytes as the main cell population involved in the gene expression changes. Finally, this study gives an insight into the molecular changes that may occur in early stages of PD and help us to understand why males are more affected by PD than females. [less ▲]

Three Essays on the General Equilibrium Effects of Human Interactions

Doctoral thesis (2023)

The overarching theme of this PhD thesis is human mobility and its externalities, particularly in the context of labour and health economics. Through rigorous modelling and analysis, the three chapters of ... [more ▼]

The overarching theme of this PhD thesis is human mobility and its externalities, particularly in the context of labour and health economics. Through rigorous modelling and analysis, the three chapters of the thesis demonstrate the potential benefits of policies that regulate human mobility. In the first chapter of my PhD, I examine how language training can improve the functioning of the labour market, with a particular focus on immigrants with high skills who face language barriers. I argue that fully funding the cost of language acquisition for migrants can bring significant benefits to the economy and migrants, but may marginally worsen the labour market performance of low-skilled natives. Using a search and matching framework with two-dimensional skill heterogeneity, I model the effects of a language acquisition subsidy on migrants' labour market integration and its impact on natives' labour market performance. My study finds that subsidizing language acquisition costs may increase the GDP of the German economy by approximately ten billion dollars by decreasing the aggregate unemployment rate and skill mismatch rate and increasing the share of job vacancies requiring high generic skills. The second chapter of my PhD explores the challenges involved in devising social contact limitation policies as a means of controlling infectious disease transmission. Using an economic-epidemiological model of COVID-19 transmission, I evaluate the effectiveness of different intervention strategies and their consequences on public health, social welfare and economic outcomes. The findings emphasize the importance of responsiveness in implementing social contact limitations, rather than solely focusing on their stringency, and suggest that early interventions lead to the lowest losses in economy and mental well-being for a given number of life losses. The study has broader implications for managing the societal impact of infectious diseases and highlights the need to continue refining our understanding of these trade-offs and developing adaptable models and policy tools to safeguard public health while minimizing social and economic consequences. Overall, the study offers a robust and versatile framework for understanding and navigating the challenges posed by public health crises and pandemics. The third chapter of my PhD builds on the economic-epidemiological model developed in Chapter 2 to analyze the multifaceted effects of vaccine hesitancy in controlling the spread of infectious diseases, with a particular focus on the COVID-19 pandemic in Belgium. The study utilizes actual vaccination rates by age group until June 2021 and simulates the following months by incorporating realistic properties such as temporary immunity, age-specific vaccination hesitancy rates, daily vaccination capacity, and vaccine efficacy rate. The baseline scenario with an overall 27.1$\%$ vaccine hesitancy rate indicates that current vaccination rates in Belgium are sufficient to control the spread of COVID-19 without imposing social contact limitations. However, hypothetical scenarios with higher disease transmission rates demonstrate the high costs of vaccine hesitancy, resulting in significant losses in labour supply, mental well-being, and life losses. Throughout this thesis, I have described the costs and benefits induced by mobility, and shown that mobility policies make winners and losers. In Chapter 1, subsidizing the cost of language acquisition for migrants can bring significant benefits to the economy and migrants, but may marginally worsen the labour market performance of low-skilled natives. In Chapter 2, stringent policies alleviate health losses, but they impact economic activity and mental health. In Chapter 3, the health externalities generated by human interactions impose a potential tradeoff between values, namely the freedom to move and the freedom to choose to get vaccinated. In each of these chapters, I quantify these tradeoffs. Another important insight from this thesis is the need to incorporate behavioural aspects into macro models evaluating the consequences of policies related to human mobility. In the thesis, these aspects include individual investments in language training, decision-making on infection avoidance, social contacts, labour supply, and vaccination decisions. can lead to more effective policies that balance the interests of various stakeholders. Overall, this thesis contributes to the literature on human mobility by highlighting the potential benefits and challenges associated with it, and the need for nuanced and responsive policymaking that takes into account behavioural aspects and externalities. The insights gained from this thesis can be relevant for future research in economics on topics related to human mobility, public health, and labour market integration. [less ▲]

Intermarriage and the Integration of Immigrants

Doctoral thesis (2023)

L’avènement du futur cadre juridique international dédié aux activités d’exploration et d’utilisation des ressources de l’espace extra-atmosphérique

Doctoral thesis (2023)

L’espace extra-atmosphérique n’est pas un espace sans droit. Toute activité d’exploration et d’utilisation de l’espace extra-atmosphérique, de la Lune et des corps célestes doit être menée conformément au ... [more ▼]

L’espace extra-atmosphérique n’est pas un espace sans droit. Toute activité d’exploration et d’utilisation de l’espace extra-atmosphérique, de la Lune et des corps célestes doit être menée conformément au droit international de l’espace. Quid des activités axées sur les ressources spatiales ? Peuvent-elles être légalement conduites dans l’espace ? Sont-elles conformes au droit international de l’espace ? Afin de répondre, entre autres, aux questions posées ci-avant, le cadre légal international qui régit actuellement les activités spatiales ainsi que trois exemples de lois nationales autorisant l’appropriation des ressources de l’espace extra-atmosphérique ont été étudiés. Par ailleurs, il a été avancé qu’afin (i) de maintenir la paix et la sécurité internationales et de favoriser la coopération et la compréhension internationales, (ii) d’assurer un niveau de sécurité juridique élevé, (iii) d’encourager le développement du secteur de l’industrie spatiale axée sur l’exploration et l’utilisation des ressources de l’espace extra-atmosphérique, l’adoption d’un cadre juridique international dédié aux activités d’exploration et d’utilisation des ressources de l’espace extra-atmosphérique était indispensable. Les recherches menées dans le cadre de la thèse ont permis de discerner et de proposer un certain nombre d’éléments et de mesures techniques et organisationnelles qui pourraient composer le futur cadre juridique international qui régira les activités axées sur les ressources spatiales. [less ▲]

THE ORAL MICROBIOME IN THE SCOPE OF EARLY LIFE ADVERSITY

Doctoral thesis (2023)

In a way, the microbiome has been on earth long before we humans appeared in any evolution theory. This ethereal “organ” has grabbed the attention of biological research and will continue to expand in the ... [more ▼]

In a way, the microbiome has been on earth long before we humans appeared in any evolution theory. This ethereal “organ” has grabbed the attention of biological research and will continue to expand in the years to come. Growing evidence supports that its presence is vital to the host and particularly to us humans. Although largest part of the mirobiome research focuses on the gut microbiome, the microbial communities actually stretch in the farthest off body sites. The last few years, oral microbiome (OM) has started to steal some light. Despite of the geolocation, the microbiome maintains an intimate relationship with its host. Our long term health appears to be closely linked and sustained by a “happy”, healthy and symbiotic microbiome.Disease and state of illness, whether it is of physiological or psychological nature or both can result from many different causes and routes. Remarkably, about a third of the adult population worldwide seems to become ill as concequence of certain difficult, stressful and intense experiences during their childhood. The field of social, biological, medical and environmental sciences describes these experiences as “Early Life Adversity (ELA)” and the marrow of the “Developmental Origins of Health and Disease (DOHaD)” . Since the 80s that this sociobiological theory was introduced and described, research revealed that ELA factors could be anything from pregnancy complications, medication, birth complications, infections, social isolation, socioeconomic status, orphanhood, toxic environment and any other event that is capable of inducing extreme stress to an individual. Concequences of ELA experiences have also been introduced as wide range of common later life diseases. An institutionalisation cohort as a model of ELA, the EPIPATH, which was initially established to study the longterm immune and cardiometabolic effect of institutionalisation was used for this Thesis. At first, I aimed to investigate what was the effect of institutionalisation on the OM of these individuals using their oral taxonomic composition. Secondly, I aimed to identify a link between this composition and the immune profiles of the participants. Thirdly, knowing the stress signatures of the cohort following pre-existing cortisol and glucose measurements, I aimed to detect plausible interaction between the microbiome and those data. Lastly, I aimed to find mechanistic evidence of our prior observed associations by looking into the metabolome of the oral microbiome. Altogether our data revealed observations of a complex system of bidirectional interactions between ELA, the OM and the immune markers of cytotoxicity and immune senescence together with a particular profile of glucose and cortisol kinetics following exposure to social stress. Besides, our data expanded to the exposure of the first mechanistic cues of ELA traces on the OM metabolome. The research conducted for this thesis brings to light important evidence on how ELA interacts with the OM leading to a certain disease phenotype. [less ▲]

Topics and institutions in the reproduction of intersectional inequalities in science

Doctoral thesis (2023)

Learning Optimisation Algorithms over Graphs

Doctoral thesis (2023)

The paradigm of learning to optimise relies on the following principle: instead of designing an algorithm to solve a problem, we design an algorithm which will automate the design of such a solver. The ... [more ▼]

The paradigm of learning to optimise relies on the following principle: instead of designing an algorithm to solve a problem, we design an algorithm which will automate the design of such a solver. The initial idea was to alleviate the limitations stated by the No Free Lunch Theorem by producing an algorithm which efficiency is less dependent upon known instances of the problem to tackle. Hyper-heuristics constitute the main learning-to-optimise techniques. These rely on a high-level algorithm performing a search process into a space of low-level heuristics to tackle a given problem. Because the latter search space is problem-dependent, the vast majority of hyper-heuristics are designed to tackle a specific problem. Due to this lack of generality, existing works fully redesign hyper-heuristics when tackling a new problem, despite the fact that they may share a similar structure. In this dissertation, we tackle this challenge by proposing a generic way for learning to optimise any problem. To this end, this thesis introduces three main contributions: (i) an analysis of the formal functioning of learning-to-optimise techniques; (ii) a model of generic hyper-heuristic, named Algorithm Learner for Graph Optimisation problems (ALGO), constituting the central point of this work; (iii) a real-world use case where we use our generic hyper-heuristic to automate the design of behaviours within a swarm of drones. In the first part, we provide a formalism for optimisation and learning concepts, which we use to describe the large body of knowledge that combines two layers of optimisation and/or learning. We then put an emphasis on approaches using learning to improve an optimisation process, i.e., aiming at learning to optimise. In the second part, we present ALGO, our model of generic hyper-heuristic. We explain how we abstract from a given problem with a graph structure so that it can be used to tackle any optimisation problem. We also detail the steps to follow in order to use ALGO to tackle a given problem. We finally present the modularity of ALGO with inner components that a user can implement. The second part ends with a validation of our model, i.e., using ALGO to tackle a classical optimisation problem. In the third part, we use ALGO to tackle the problem of area surveillance with a swarm of drones. We demonstrate that ALGO constitutes a novel and efficient way to automate the design of such a distributed and multi-objective problem. [less ▲]

Daz was ein dinc, daz hiez… buoch-Literarisierungen in mittelhochdeutschen Texten des späten 12. und 13. Jahrhunderts

Doctoral thesis (2023)

Security and Privacy of Resource Constrained Devices

Doctoral thesis (2023)

The thesis aims to present a comprehensive and holistic overview on cybersecurity and privacy & data protection aspects related to IoT resource-constrained devices. Chapter 1 introduces the current ... [more ▼]

The thesis aims to present a comprehensive and holistic overview on cybersecurity and privacy & data protection aspects related to IoT resource-constrained devices. Chapter 1 introduces the current technical landscape by providing a working definition and architecture taxonomy of ‘Internet of Things’ and ‘resource-constrained devices’, coupled with a threat landscape where each specific attack is linked to a layer of the taxonomy. Chapter 2 lays down the theoretical foundations for an interdisciplinary approach and a unified, holistic vision of cybersecurity, safety and privacy justified by the ‘IoT revolution’ through the so-called infraethical perspective. Chapter 3 investigates whether and to what extent the fast-evolving European cybersecurity regulatory framework addresses the security challenges brought about by the IoT by allocating legal responsibilities to the right parties. Chapters 4 and 5 focus, on the other hand, on ‘privacy’ understood by proxy as to include EU data protection. In particular, Chapter 4 addresses three legal challenges brought about by the ubiquitous IoT data and metadata processing to EU privacy and data protection legal frameworks i.e., the ePrivacy Directive and the GDPR. Chapter 5 casts light on the risk management tool enshrined in EU data protection law, that is, Data Protection Impact Assessment (DPIA) and proposes an original DPIA methodology for connected devices, building on the CNIL (French data protection authority) model. [less ▲]

A Principled Judicial-Regulatory Framework Analysis of Approaches to Hate Speech under European, German, Irish and English Human Rights Law

Doctoral thesis (2023)

TARGETED, REALISTIC AND NATURAL FAULT INJECTION : (USING BUG REPORTS AND GENERATIVE LANGUAGE MODELS)

Doctoral thesis (2023)

Artificial faults have been proven useful to ensure software quality, enabling the simulation of its behaviour in erroneous situations, and thereby evaluating its robustness and its impact on the ... [more ▼]

Artificial faults have been proven useful to ensure software quality, enabling the simulation of its behaviour in erroneous situations, and thereby evaluating its robustness and its impact on the surrounding components in the presence of faults. Similarly, by introducing these faults in the testing phase, they can serve as a proxy to measure the fault revelation and thoroughness of current test suites, and provide developers with testing objectives, as writing tests to detect them helps reveal and prevent eventual similar real ones. This approach – mutation testing – has gained increasing fame and interest among researchers and practitioners since its appearance in the 1970s, and operates typically by introducing small syntactic transformations (using mutation operators) to the target program, aiming at producing multiple faulty versions of it (mutants). These operators are generally created based on the grammar rules of the target programming language and then tuned through empirical studies in order to reduce the redundancy and noise among the induced mutants. Having limited knowledge of the program context or the relevant locations to mutate, these patterns are applied in a brute-force manner on the full code base of the program, producing numerous mutants and overwhelming the developers with a costly overhead of test executions and mutants analysis efforts. For this reason, although proven useful in multiple software engineering applications, the adoption of mutation testing remains limited in practice. Another key challenge of mutation testing is the misrepresentation of real bugs by the induced artificial faults. Indeed, this can make the results of any relying application questionable or inaccurate. To tackle this challenge, researchers have proposed new fault-seeding techniques that aim at mimicking real faults. To achieve this, they suggest leveraging the knowledge base of previous faults to inject new ones. Although these techniques produce promising results, they do not solve the high-cost issue or even exacerbate it by generating more mutants with their extended patterns set. Along the same lines of research, we start addressing the aforementioned challenges – regarding the cost of the injection campaign and the representativeness of the artificial faults – by proposing IBIR; a targeted fault injection which aims at mimicking real faulty behaviours. To do so, IBIR uses information retrieved from bug reports (to select relevant code locations to mutate) and fault patterns created by inverting fix patterns, which have been introduced and tuned based on real bug fixes mined from different repositories. We implemented this approach, and showed that it outperforms the fault injection performed by traditional mutation testing in terms of semantic similarity with the originally targeted fault (described in the bug report), when applied at either project or class levels of granularity, and provides better, statistically significant, estimations of test effectiveness (fault detection). Additionally, when injecting only 10 faults, IBIR couples with more real bugs than mutation testing even when injecting 1000 faults. Although effective in emulating real faults, IBIR’s approach depends strongly on the quality and existence of bug reports, which when absent can reduce its performance to that of traditional mutation testing approaches. In the absence of such prior and with the same objective of injecting few relevant faults, we suggest accounting for the project’s context and the actual developer’s code distribution to generate more “natural” mutants, in a sense where they are understandable and more likely to occur. To this end, we propose the usage of code from real programs as a knowledge base to inject faults instead of the language grammar or previous bugs knowledge, such as bug reports and bug fixes. Particularly, we leverage the code knowledge and capability of pre-trained generative language models (i.e. CodeBERT) in capturing the code context and predicting developer-like code alternatives, to produce few faults in diverse locations of the input program. This way the approach development and maintenance does not require any major effort, such as creating or inferring fault patterns or training a model to learn how to inject faults. In fact, to inject relevant faults in a given program, our approach masks tokens (one at a time) from its code base and uses the model to predict them, then considers the inaccurate predictions as probable developer-like mistakes, forming the output mutants set. Our results show that these mutants induce test suites with higher fault detection capability, in terms of effectiveness and cost-efficiency than conventional mutation testing. Next, we turn our interest to the code comprehension of pre-trained language models, particularly their capability in capturing the naturalness aspect of code. This measure has been proven very useful to distinguish unusual code which can be a symptom of code smell, low readability, bugginess, bug-proneness, etc, thereby indicating relevant locations requiring prior attention from developers. Code naturalness is typically predicted using statistical language models like n-gram, to approximate how surprising a piece of code is, based on the fact that code, in small snippets, is repetitive. Although powerful, training such models on a large code corpus can be tedious, time-consuming and sensitive to code patterns (and practices) encountered during training. Consequently, these models are often trained on a small corpus and thus only estimate the language naturalness relative to a specific style of programming or type of project. To overcome these issues, we propose the use of pre-trained generative language models to infer code naturalness. Thus, we suggest inferring naturalness by masking (omitting) code tokens, one at a time, of code sequences, and checking the models’ ability to predict them. We implement this workflow, named CodeBERT-NT, and evaluate its capability to prioritize buggy lines over non-buggy ones when ranking code based on its naturalness. Our results show that our approach outperforms both, random-uniform- and complexity-based ranking techniques, and yields comparable results to the n-gram models, although trained in an intra-project fashion. Finally, We provide the implementation of tools and libraries enabling the code naturalness measuring and fault injection by the different approaches and provide the required resources to compare their effectiveness in emulating real faults and guiding the testing towards higher fault detection techniques. This includes the source code of our proposed approaches and replication packages of our conducted studies. [less ▲]

The Application of the European Account Preservation Order in Germany, Luxembourg and Spain: A Comparative-Empirical Analysis

Doctoral thesis (2023)

Regulation (EU) 655/2014, establishing a European Account Preservation Order (‘EAPO Regulation’), introduced the very cross-border interim measure at the European Union level. As its name indicates, it ... [more ▼]

Regulation (EU) 655/2014, establishing a European Account Preservation Order (‘EAPO Regulation’), introduced the very cross-border interim measure at the European Union level. As its name indicates, it permits the direct cross-border attachment of funds in bank accounts. Furthermore, it contains a specific mechanism to search for the debtors’ bank accounts that does not exist in all EU-Member States. Although the EAPO is a self-standing procedure, it combines uniform standards with other aspects of the procedure that depend on national law. This dissertation studies and compares the application of the EAPO Regulation in three jurisdictions: Germany, Luxembourg, and Spain. It aims to understand the incorporation of the EAPO procedure within a national civil procedural system and the impact national law has on the EAPO procedure as a whole. The comparative approach determines the different ways Member States apply the EAPO procedures. This comparative analysis follows both a theoretical and empirical approach. The empirical approach relies on qualitative and quantitative data on the functioning of the EAPO obtained from stakeholders, case law databases, and institutional statistical data. The empirical side of the research seeks to identify specific issues courts and practitioners encounter in real-life EAPO cases, and whether such issues are the same or different depending on the jurisdiction where the EAPO is applied. Relying on the outcome of the comparative-empirical analysis, specific policy-making recommendations are designed and intended to improve the application of the EAPO at the EU and the national level. [less ▲]

Normative coherence and Antifragility for the Safe and Just space: a new pathway to crisis response and sustainability

Doctoral thesis (2023)

DATA-DRIVEN DEFEATURING OF COMPLEX GEOMETRICAL STRUCTURES

Doctoral thesis (2023)

This work is part of the reconstruction of data from images. Its purpose is to develop methods to generate a surface of CAO type (B-Spline or NURBS). Indeed, obtaining a mathematical representation of the ... [more ▼]

This work is part of the reconstruction of data from images. Its purpose is to develop methods to generate a surface of CAO type (B-Spline or NURBS). Indeed, obtaining a mathematical representation of the surface of a solid body from point clouds, images or tetrahedral meshes is a fundamental task of 21st century digital engineering, where simulators interact with real systems. In this thesis, we will develop new algorithms for reconstruction of CAO geometry. First, in order to determine a NURBS surface, a control network, i.e. a quadrangular mesh, is required. Using the eigenfunctions of a Graph Laplacian problem and thanks to the discrete Morse theory, a control network is determined. The surface obtained using this mesh is not a priori optimal, which is why an optimization algorithm is introduced. It allows to adjust the NURBS surface to the triangulation and thus to best approximate the geometry of the object. Then, a model selection is carried out. To do so, a regression model is set up to compare the surfaces obtained with 3D images, and a surface is chosen using a information criterion. These steps being established, we will no longer consider only the noise of the data, but also that of the solution. Thus, using a sampling method, a probabilistic distribution of surface is determined. Finally, in perspective, constraints are applied to the graph Laplacian problem in order to align the NURBS patches along a given curve, for example in the case of an object with a marked edge. The methods developed are robust and do not depend on the topology of the desired 3D object, that is to say that the algorithm works on a wide range of shapes. We apply the developed methodologies in the biomedical field, with examples of vertebrae and femurs. This would make it possible to have the scanned object, a bone, and the implant in the same "format" and thus the adjustment of the implant will be carried out more easily. [less ▲]

Spaceborne GNSS reflectometry for land remote sensing studies

Doctoral thesis (2023)

Understanding, quantifying and monitoring soil moisture is important for many applications, e.g., agriculture, weather forecasting, occurrence of heatwaves, droughts and floods, and human health. At a ... [more ▼]

Understanding, quantifying and monitoring soil moisture is important for many applications, e.g., agriculture, weather forecasting, occurrence of heatwaves, droughts and floods, and human health. At a large scale, satellite microwave remote sensing has been used to retrieve soil moisture information. Surface water has also been detected and monitored through remote sensing orbital platforms equipped with passive microwave, radar, and optical sensors. The use of reflected L-band Global Navigation Satellite System (GNSS) signals represents an emerging remote sensing concept to retrieve geophysical parameters. In GNSS Reflectometry (GNSS-R) these signals are repurposed to infer properties of the surface from which they reflect as they are sensitive to variations in biogeophysical parameters. NASA's Cyclone GNSS (CYGNSS) is the first mission fully dedicated to spaceborne GNSS-R. The eight-satellite constellation measures Global Positioning System (GPS) reflected L1 (1575.42 MHz) signals. Spire Global, Inc. has also started developing their GNSS-R mission, with four satellites currently in orbit. In this thesis we propose and validate a method to retrieve large-scale near-surface soil moisture and a method to map and monitor inundations using spaceborne GNSS-R. Our soil moisture model is based on the assumption that variations in surface reflectivity are linearly related to variations in soil moisture and uses a new method to normalize the observations with respect to the angle of incidence. The normalization method accounts for the spatially varying effects of coherent and incoherent scattering. We found a median unbiased root-mean-square error (ubRMSE) of 0.042 cm3 cm-3 when comparing our method to two years of Soil Moisture Active Passive (SMAP) data and a median ubRMSE of 0.059 cm3 cm-3 compared to the observations of 207 in-situ stations. Our results also showed an improved temporal resolution compared to sensors traditionally used for this purpose. Assessing Spire and CYGNSS data over a region in south east Australia, we observed a similar behavior in terms of surface reflectivity and sensitivity to soil moisture. As Spire satellites collect data from multiple GNSS constellations, we found that it is important to differentiate the observations when calibrating a soil moisture model. The inundation mapping method that we propose is based on a track-wise approach. When classifying the reflections track by track the influence of the angle of incidence and the GNSS transmitted power are minimized or eliminated. With CYGNSS data we produced more than four years of monthly surface water maps over the Amazon River basin and the Pantanal wetlands complex with a spatial resolution of 4.5 km. With GNSS-R we could overcome some of the limitations of optical and microwave remote sensing methods for inundation mapping. We used a set of metrics commonly used to evaluate classification performance to assess our product and discussed the differences and similarities with other products. [less ▲]

STUDY OF EARLY MELANOMA BRAIN METASTASIS MECHANISMS USING IN VITRO AND IN VIVO MODELS OF TUMOR INVASION

Doctoral thesis (2023)

Of all skin cancers, melanoma is the most fatal. Of all cancer types, melanoma is also the cancer with the highest level of brain tropism. Approximately 50% of patients with stage IV melanoma are ... [more ▼]

Of all skin cancers, melanoma is the most fatal. Of all cancer types, melanoma is also the cancer with the highest level of brain tropism. Approximately 50% of patients with stage IV melanoma are diagnosed with melanoma brain metastases. A percentage that rises when postmortem patients are also taken into account. Following lung cancer and breast cancer, melanoma is the leading cause of malignant metastasis to the central nervous system. Of all metastatic brain tumors, melanoma represents 6-12% of cases. The overall survival rate following a diagnosis of melanoma brain metastases has been historically low. However, over the past ten years, advances in targeted therapies as well as in immunotherapies have significantly improved the survival rate of patients with advanced melanoma. Melanoma brain metastases most frequently occur at the junction between the gray and the white matter and in the frontal lobe. In order to reach the brain parenchyma, metastases must cross the brain vasculature. The specific properties of the blood vessels that perfuse the central nervous system are referred to as the blood-brain barrier. They allow these vessels to finely regulate the flow of cells, ions and molecules between the bloodstream and the brain parenchyma in order to preserve brain homeostasis for the proper functioning of neurons and the protection of the brain against toxic and pathogenic agents. Abnormalities in this functional interfacing barrier that separates the brain from the bloodstream are a critical element in the development and progression of several neurological pathologies. A poor understanding of the early mechanisms of metastasis crossing the blood-brain barrier constitutes an obstacle to the development of effective preventive therapeutic strategies as well as a particularly challenging domain of interest as it is one of the most crucial and least documented steps in the metastasizing process to the brain. Here, we focused on the ideation and consequent creation of effective in vitro and in vivo models to help identify and characterize as meticulously as possible, the players that are implicated in the crossing of melanoma metastases through the blood-brain barrier to reach the brain parenchyma. We used human immortalized cells (endothelial cells, pericytes and astrocytes) in triple coculture to recreate a blood-brain barrier in vitro and be able to investigate eventual changes in the gene expression of the tumor cells crossing the model. In parallel, we have set up an in vivo murine model to recreate the process of brain metastasis by injecting melanoma tumor cells into the left ventricle of the heart and thus be able to study the early stages of blood-brain barrier invasion. The analysis of the murine tissues was performed by Correlative light-electron microscopy (CLEM) and the results obtained revealed the presence of cells in the brain that present artifacts that have the same appearance as melanosomes. Experiments using focused ion beam scanning electron microscopes (FIB-SEM) as well as nanoscale secondary ion mass spectrometry (NanoSIMS) may be conducted to take the investigation further. [less ▲]

Altered Images: A Comparative Study of Medical Portraits by Henry Tonks and Raphaël Freida in the Great War

Doctoral thesis (2023)

Estimating terrestrial water storage variations from GNSS vertical displacements in the Island of Haiti

Doctoral thesis (2023)

In the field of hydro-geodesy, ill-posed inverse problems are very common. Those problems need to be regularized to find a stabilized solution. Usually, to solve those problems, two regularization methods ... [more ▼]

In the field of hydro-geodesy, ill-posed inverse problems are very common. Those problems need to be regularized to find a stabilized solution. Usually, to solve those problems, two regularization methods are often used, Tikhonov’s regularization and Truncated Singular Value Decomposition (TSVD), with some common regularization parameter choice methods such as L-curve or General Cross Validation (GCV). This study aims to test the capacity of the Least Squares Collocation (LSC) method to estimate the terrestrial water storage variations as an original approach. First, for the forward model, we calculated the hydrological crustal loading deformation in the island of Haiti by convolving Farrell (1972) Green's function with the surface mass loading from the Global Land Data Assimilation (GLDAS). After, a dense synthetic Global Navigation Satellite System (GNSS) network is used with the LSC method to estimate the Terrestrial Water Storage (TWS) variations for the inverse problem. LSC is a natural way to stabilize an ill-posed inverse problem. Unlike Tikhonov’s or TSVD regularization method, LSC allows us to stabilize the inverse problem by including more physical information. The latter is introduced through a covariance function characterizing the observations, the parameters, and the functional link between them. One of the advantages of the LSC method is that it does not require any regularization parameter. First, we showed that, for the island of Haiti, the near field can extend until 24° around a GNSS station. Secondly, we proved that the hydrology-induced vertical deformation is part of the GNSS vertical displacement over the island. Finally, we demonstrated that the LSC may be used as a method to estimate TWS variations in dense GNSS network area. [less ▲]

New insights into plant hydraulics by experimentally pulling instead of pushing water through plant segments

Doctoral thesis (2023)

Quantifying the abuse potentiality in the investor treaty protection regime (ICSID Convention): the investment fund case study

Doctoral thesis (2023)

Le processus scriptural nothombien : une nouvelle expérience esthétique. Faire entendre ce qui appartient au silence.

Doctoral thesis (2023)

Amélie Nothomb est présente sur la scène littéraire depuis 31 ans et pourtant il y a en France un vide dans la recherche universitaire concernant son oeuvre. Cette thèse aborde la place encore et toujours ... [more ▼]

Amélie Nothomb est présente sur la scène littéraire depuis 31 ans et pourtant il y a en France un vide dans la recherche universitaire concernant son oeuvre. Cette thèse aborde la place encore et toujours problématique réservée aux autrices dans le champ et la critique littéraires français, les obstacles posés à leur légitimation et à leur consécration du fait de leur genre, obstacles renforcés si elles appartiennent non au centre mais aux marges de la francophonie, et plus encore si leur œuvre obtient un succès commercial indéniable, avec pour effet, auprès de la critique, que la légitimité et la littérarité de leurs ouvrages se voient remises en doute lorsqu’elles ne sont pas tout simplement niées. Amélie Nothomb fait donc l'objet d'une triple marginalité qui est renforcée par la difficulté de classifier son oeuvre. [less ▲]

Towards a Computational Model of General Cognitive Control Using Artificial Intelligence, Experimental Psychology and Cognitive Neuroscience

Doctoral thesis (2023)

Cognitive control is essential to human cognitive functioning as it allows us to adapt and respond to a wide range of situations and environments. The possibility to enhance cognitive control in a way ... [more ▼]

Cognitive control is essential to human cognitive functioning as it allows us to adapt and respond to a wide range of situations and environments. The possibility to enhance cognitive control in a way that transfers to real life situations could greatly benefit individuals and society. However, the lack of a formal, quantitative definition of cognitive control has limited progress in developing effective cognitive control training programs. To address this issue, the first part of the thesis focuses on gaining clarity on what cognitive control is and how to measure it. This is accomplished through a large-scale text analysis that integrates cognitive control tasks and related constructs into a cohesive knowledge graph. This knowledge graph provides a more quantitative definition of cognitive control based on previous research, which can be used to guide future research. The second part of the thesis aims at furthering a computational understanding of cognitive control, in particular to study what features of the task (i.e., the environment) and what features of the cognitive system (i.e., the agent) determine cognitive control, its functioning, and generalization. The thesis first presents CogEnv, a virtual cognitive assessment environment where artificial agents (e.g., reinforcement learning agents) can be directly compared to humans in a variety of cognitive tests. It then presents CogPonder, a novel computational method for general cognitive control that is relevant for research on both humans and artificial agents. The proposed framework is a flexible, differentiable end-to-end deep learning model that separates the act of control from the controlled act, and can be trained to perform the same cognitive tests that are used in cognitive psychology to assess humans. Together, the proposed cognitive environment and agent architecture offer unique new opportunities to enable and accelerate the study of human and artificial agents in an interoperable framework. Research on training cognition with complex tasks, such as video games, may benefit from and contribute to the broad view of cognitive control. The final part of the thesis presents a profile of cognitive control and its generalization based on cognitive training studies, in particular how it may be improved by using action video game training. More specifically, we contrasted the brain connectivity profiles of people that are either habitual action video game players or do not play video games at all. We focused in particular on brain networks that have been associated with cognitive control. Our results show that cognitive control emerges from a distributed set of brain networks rather than individual specialized brain networks, supporting the view that action video gaming may have a broad, general impact of cognitive control. These results also have practical value for cognitive scientists studying cognitive control, as they imply that action video game training may offer new ways to test cognitive control theories in a causal way. Taken together, the current work explores a variety of approaches from within cognitive science disciplines to contribute in novel ways to the fascinating and long tradition of research on cognitive control. In the age of ubiquitous computing and large datasets, bridging the gap between behavior, brain, and computation has the potential to fundamentally transform our understanding of the human mind and inspire the development of intelligent artificial agents. [less ▲]

Reading aloud practices: Providing joint accessibility to text within an unfamiliar interface-mediated game activity

Doctoral thesis (2023)

This work examines the practice of reading aloud in the interactional context of adult participants engaging in an interface-mediated collaborative game activity. With a conversation analytic approach ... [more ▼]

This work examines the practice of reading aloud in the interactional context of adult participants engaging in an interface-mediated collaborative game activity. With a conversation analytic approach onto video data of user studies, empirical cases of reading aloud are presented. It is shown how participants multimodally co-organise reading aloud in-interaction for providing accessibility to game text in a game that is unfamiliar to them. With reading aloud, participants meet the interactional challenge of making game text audibly accessible that is not always visually accessible for all participant alike. This practice is not only conducted for another but with another in a truly joint fashion, working as a continuer to accomplish the unfamiliar game. [less ▲]

Soil viral particles as tracers of surface water sources and flow paths

Doctoral thesis (2023)

Albeit recent technological developments (e.g. field deployable instruments operating at high temporal frequencies), experimental hydrology is a discipline that remains measurement limited. From this ... [more ▼]

Albeit recent technological developments (e.g. field deployable instruments operating at high temporal frequencies), experimental hydrology is a discipline that remains measurement limited. From this perspective, trans-disciplinary approaches may create valuable opportunities to enlarge the number of tools available for investigating hydrological processes. Tracing experiments are usually performed in order to investigate the water flow pathways and water sources in underground areas. Since the 19th century, researchers have worked with hydrological tracers to do this. Among them, the fluorescent dyes and the isotopes are the most commonly used to follow the water flow while others like salts or bacteriophages are employed as additional tracers to those mentioned above. Bacteriophages are the least known of all, but it has been studied since the 1960s as hydrological tracers, especially in karstic environments. The purpose is to evaluate the potential for bacteriophages naturally occurring in soils to serve as a new environmental tracer of hydrological processes. We hypothesize that such viral particles can be a promising tool in water tracing experiments since they are safe for ecosystems. In both fields of hydrology and virology, the knowledge regarding the fate of bacteriophages within the pedosphere is still limited. Their study would not only allow proposing potential new candidates to enlarge the hydrological tracers available, but also improving the current knowledge about the bacteriophage communities in soil and their interactions with certain environmental factors. For this purpose, we aim at describing the bacteriophage communities occurring in the soil through shotgun metagenomics analysis. Those viruses are widely spread in the pedosphere, and we assume that they have specific signatures according to the type of soil. Then, bacteriophage populations will be investigated in the soil water to analyse the dis/similarities between the two communities as well as their dynamics in the function of the precipitation events. This way, with a relatively high abundance of soil and soil water and a capacity of being mobilised, good bacteriophage candidates could be selected as hydrological tracers. [less ▲]

Robust estimation in exponential families: from theory to practice

Doctoral thesis (2023)

REMOTE PLASMA CHEMICAL VAPOUR DEPOSITION FOR GAS DIFFUSION LAYER AND PROTON EXCHANGE MEMBRANE SYNTHESIS FOR FUEL CELLS

Doctoral thesis (2023)

Climate change due to the increase in GHG emissions and energy crisis due to scarcity of fossil fuel availability are an ever growing issue for the planet and countries. The decarbonization and the ... [more ▼]

Climate change due to the increase in GHG emissions and energy crisis due to scarcity of fossil fuel availability are an ever growing issue for the planet and countries. The decarbonization and the sustainability of the energy sector is one of the top priority to achieve a resilient system. Hydrogen has been considered for decades to be used as an alternative for fossil fuel and now is the time of development for an hydrogen based economy. Fuel cells are devices that convert the hydrogen chemical energy into electrical energy and is one the main component considered for the hydrogen economy. However, much is yet to be achieved to make their manufacturing as cheap and as efficient as possible. Chemical vapour deposition (CVD) is a technique used to synthesize solid materials from gaseous precursors which has the advantages, over wet chemistry, to reduce wastes of production, to be cheap, to make pure solid materials and to be easily scalable. In this thesis we investigated the possibility to use CVD to produce two major components of fuel cells, namely the gas diffusion layer and the proton exchange membrane. The results were highly promising regarding the elaboration of gas diffusion layers and a CVD prototype was assembled to make the highly complex copolymerization of proton exchange membrane a reality with promising initial results. [less ▲]

Analyzing the Unanalyzable: an Application to Android Apps

Doctoral thesis (2023)

In general, software is unreliable. Its behavior can deviate from users’ expectations because of bugs, vulnerabilities, or even malicious code. Manually vetting software is a challenging, tedious, and ... [more ▼]

In general, software is unreliable. Its behavior can deviate from users’ expectations because of bugs, vulnerabilities, or even malicious code. Manually vetting software is a challenging, tedious, and highly-costly task that does not scale. To alleviate excessive costs and analysts’ burdens, automated static analysis techniques have been proposed by both the research and practitioner communities making static analysis a central topic in software engineering. In the meantime, mobile apps have considerably grown in importance. Today, most humans carry software in their pockets, with the Android operating system leading the market. Millions of apps have been proposed to the public so far, targeting a wide range of activities such as games, health, banking, GPS, etc. Hence, Android apps collect and manipulate a considerable amount of sensitive information, which puts users’ security and privacy at risk. Consequently, it is paramount to ensure that apps distributed through public channels (e.g., the Google Play) are free from malicious code. Hence, the research and practitioner communities have put much effort into devising new automated techniques to vet Android apps against malicious activities over the last decade. Analyzing Android apps is, however, challenging. On the one hand, the Android framework proposes constructs that can be used to evade dynamic analysis by triggering the malicious code only under certain circumstances, e.g., if the device is not an emulator and is currently connected to power. Hence, dynamic analyses can -easily- be fooled by malicious developers by making some code fragments difficult to reach. On the other hand, static analyses are challenged by Android-specific constructs that limit the coverage of off-the-shell static analyzers. The research community has already addressed some of these constructs, including inter-component communication or lifecycle methods. However, other constructs, such as implicit calls (i.e., when the Android framework asynchronously triggers a method in the app code), make some app code fragments unreachable to the static analyzers, while these fragments are executed when the app is run. Altogether, many apps’ code parts are unanalyzable: they are either not reachable by dynamic analyses or not covered by static analyzers. In this manuscript, we describe our contributions to the research effort from two angles: ① statically detecting malicious code that is difficult to access to dynamic analyzers because they are triggered under specific circumstances; and ② statically analyzing code not accessible to existing static analyzers to improve the comprehensiveness of app analyses. More precisely, in Part I, we first present a replication study of a state-of-the-art static logic bomb detector to better show its limitations. We then introduce a novel hybrid approach for detecting suspicious hidden sensitive operations towards triaging logic bombs. We finally detail the construction of a dataset of Android apps automatically infected with logic bombs. In Part II, we present our work to improve the comprehensiveness of Android apps’ static analysis. More specifically, we first show how we contributed to account for atypical inter-component communication in Android apps. Then, we present a novel approach to unify both the bytecode and native in Android apps to account for the multi-language trend in app development. Finally, we present our work to resolve conditional implicit calls in Android apps to improve static and dynamic analyzers. [less ▲]