Excellent Doctoral Thesis Awards 2023 in science

The Doctoral School of Science and Engineering (DSSE) awarded 10 doctoral candidates for their outstanding doctoral thesis at the University of Luxembourg. In addition, the Fondation Laval gave an award for the best thesis in electricity. Their research covers a wide variety of topics, tackling important issues and providing innovative solutions. They now continue their career in both academia and industry.

Doctoral Programme in Complex Systems Science

In modern engineering applications, high-fidelity computational models are often impractical due to their slow performance and lack of certainty in predictions. Saurabh’s thesis overcomes these limitations by introducing innovative deep learning surrogate frameworks that are scalable, robust, require minimal hyper-parameter tuning, and exhibit fast performance at the inference stage. These surrogate frameworks are constructed using various deep learning techniques under deterministic as well as Bayesian settings, and they are employed to solve problems in computational mechanics. Additionally, the thesis introduces two novel deep learning layers that can be useful for a wide range of engineering applications.

—

From population collapses to cell-fate decision, critical phenomena are abundant in complex real-world systems. Among modelling theories to address them, the critical transitions framework gained traction for its purpose of determining classes of critical mechanisms and identifying generic indicators to detect and alert them (“early warning signals”). This thesis contributes to such research field by elucidating its relevance within the systems biology landscape, by providing a systematic classification of leading mechanisms for critical transitions, and by assessing the theoretical and empirical performance of early warning signals. The thesis thus bridges general results concerning the critical transitions field – possibly applicable to multidisciplinary contexts – and specific applications in biology and epidemiology, towards the development of sound risk monitoring system.

Doctoral Programme in Computer Science and Computer Engineering

In his thesis, Gabriel focuses on novel signal processing techniques for enabling the use of radar devices for contactless monitoring vital signs. Vital sign information provides valuable insight into a patient’s condition, including how well they are responding to medical treatment. However, conventional contact-based devices using adhesives and wires are inappropriate for long-term continuous monitoring. Besides mobility restrictions and stress, they can cause epidermal damage and even lead to pressure necrosis. This is especially important for specific groups of individuals such as neonates, elderly, and burned patients. In collaboration with IEE S.A. and the Department of General Pediatrics and Neonatology from the Saarland University Medical School, the proposed radar solution was already successfully tested for monitoring the respiration of premature babies in a neonatal intensive care unit (see article).

—

In recent years, there has been a concerning increase in malicious apps infiltrating official app markets like the Google Play store. These apps manage to bypass security measures, endangering users’ devices and personal information. These malware affect thousands, even millions, of users, highting the importance of robust security measures for mobile devices and app markets. In his research, Jordan has been working on developing effective methods to detect and prevent stealthy malicious behavior in mobile apps. His research involves innovative approaches and tools to address real-world software engineering and security challenges. For instance, thanks to his research, he has identified new malware in the Google Play, illustrating his dedication to improving mobile device security. Hence, his research has the potential to affect millions of users worldwide.

—

This thesis aims to look into the engineering aspects of next-generation cryptography, i.e., the problems concerning implementation efficiency and security. Hao first explored efficient software implementation approaches for lattice-based PQC on constrained devices. Then, he studied how to speed up isogeny-based PQC on modern high-performance processors especially by using their powerful vector units. Moreover, he researched how to design sophisticated yet low-area instruction set extensions to further accelerate software implementations of LWC and long-integer-arithmetic-based PQC. Finally, to address the threats from potential power side-channel attacks, he presented a concept of using special leakage-aware instructions to eliminate overwriting leakage for masked software implementations.

Doctoral Programme in Engineering Sciences

Automating assembly processes in High-Mix, Low Volume (HMLV) manufacturing remains challenging, especially for Small and Medium-sized Enterprises (SMEs). Consequently, many companies still rely on a significant amount of manual operations with a low degree of automation. In his research, Alexej developed a multidisciplinary approach to leverage learning-based Computer Vision (CV) methods to enable the automation of assembly processes in SMEs operating in an HMLV environment. With the proposed procedure, it is possible to identify process-relevant parameters critical for automation and determine them via learning-based CV models. Following the procedure, several CV models are developed for a combined pick & place and gluing process and implemented on a technology demonstrator reaching Technology Readiness Level 4. In other use cases, the developed models indicate high performance, robustness, and flexibility and mark the baseline for process automation. The application in an industrial context can lead to increased productivity, higher quality, and reduced rework/scrap, securing the competitiveness of SMEs in a global market.

Doctoral Programme in Mathematics and Applications

Her PhD project focuses on robust estimation within the one-parameter exponential family. Rho-estimation, a novel general approach, is specifically designed to address scenarios involving contaminated data, misspecified models, and/or the presence of outliers. Her proposed methodology, based on rho-estimation, encompasses three distinct estimation strategies that effectively tackle these challenges. Under certain conditions, several theoretical results have been extended from the Gaussian case to all one-parameter exponential families, thereby contributing to the theoretical gap in the literature. Additionally, she explored various applications, such as estimation using deep neural networks and change-point detection. From a practical standpoint, she investigated the algorithm to search for the rho-estimator. It turns out that the resulting estimator achieves optimality and robustness performance under different settings.

Doctoral Programme in Physics and Materials Science

Inspired by circular economy principles, my thesis focuses on the development of renewable and recyclable polybenzoxazines, a specific type of plastics that are known for their excellent thermo-mechanical properties. The renewability aspect was addressed using various bio-based precursors, with the ultimate objective of preparing material from lignin, a natural waste product. Polybenzoxazine vitrimer precursors were designed with respect to the principles of Green Chemistry. Polybenzoxazine vitrimer materials exhibit self-healing, remoldability, and chemical and mechanical reprocessability aptitudes. This research lays the groundwork for making more sustainable materials, helping the environment by reducing plastic pollution, and paving the way for a greener future.

Doctoral Programme in Systems and Molecular Biomedicine

Resistance to treatment in cancer is common and leads to tumor relapse. Tumors are able to escape treatments due to their plasticity. As part of this plasticity, cellular metabolism is highly adaptable to cope with changing microenvironments leading to therapy resistance. Therefore, profiling cancer cells under metabolic stress offers therapeutic potential. During his PhD, Mohaned studied the metabolic potential of cancer cells when faced with limited glucose supply, a physiological relevant feature of tumors. He was able to identify a novel block of glycolysis at pyruvate kinase M2 isozyme (PKM2) and importantly he demonstrated that this block is independent of previously known mechanisms and relies on formate overflow. His thesis results are challenging the current dogma on PKM2 function and extend the current knowledge on how cancer cells cope with nutrient limitation.

—

The human gut microbiome has been shown to play an important role in the development and progression of colorectal cancer. In this project, the effects of the ketogenic diet, a high-fat low-carbohydrate diet, on the gut microbiome and on colorectal cancer are examined through the use of complex in vitro and in vivo models. The ketogenic diet was found to exhibit tumor-suppressing properties though modulation of the gut microbial composition and function, demonstrating the causal role of the gut microbiome in mediating these beneficial dietary effects. Bacterial long-chain fatty acid metabolism was found to be particularly affected by the ketogenic diet, and a mediator of its anti-cancer properties. This research paves the way for the establishment of dietary guidelines for cancer patients and for the development of microbiome-based therapies.

—

Laval prize

This thesis delves into the fabrication and study of technology-relevant functional thin films using cost-effective solution deposition methods. In his research, Alfredo focused on the ferroelectric perovskite bismuth ferrite. He investigated its potential to convert light into electricity via the bulk photovoltaic effect and to manipulate light propagation using electric fields through the electro-optic effect. In particular, he aimed to understand the relationship between the material’s structure (including its chemical composition, defects, and strain) and its photovoltaic and electro-optic properties. The findings presented in this thesis not only demonstrate new routes for fabricating low-cost polycrystalline bismuth ferrite films but also highlight their potential in advanced light-driven applications such as electro-optic modulation, light-driven actuators, and holographic data storage.