Doctoral School

Reading date: 03/02/2026

• NJOKU, UCHECHUKWU FORTUNE: Towards Effective and Interpretable Many-Objective Feature Selection in Machine Learning
  
  Author: NJOKU, UCHECHUKWU FORTUNE
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: DOCTORAL DEGREE IN COMPUTING
  Department: Department of Computer Science (CS)
  Mode: Change of supervisor
  Deposit date: 04/12/2025
  Reading date: 03/02/2026
  Reading time: 14:00
  Reading place: Sala d'Actes Marti Recober de la FIBEnllaç: https://teams.microsoft.com/dl/launcher/launcher.html?url=%2F_%23%2Fl%2Fmeetup-join%2F19%3Ameeting_NGFmN
  Thesis director: ABELLO GAMAZO, ALBERTO | BILALLI, BESIM | BONTEMPI, GIANLUCA
  Thesis abstract: Effective Machine Learning (ML) requires more than just accurate models; it also demands consideration of factors such as model complexity, fairness, and other task-specific requirements. Fulfilling these requirements begins at the data level by selecting features that con-tribute to addressing these concerns. This can benefit from a many-objective optimization approach to Feature Selection (FS).This thesis, therefore, studies Many-Objective Feature Selection (MOFS) and contributes to the development of efficient and responsible ML solutions. However, due to the large number of MOFS solutions, it comes with an interpretability challenge. Therefore, we also aim to propose a methodology for tackling this limitation of MOFS.Although FS has been long researched, previous work (on both filter and wrapper methods) has failed to address this gap by focusing only on one or at most two objectives. Also for the interpretability of FS results, no methodological approach has been proposed and rather a basic tabular representation has been used.We propose a framework that uses non-dominated sorting genetic algorithms to balance important and often conflicting objectives for FS. In particular, more than four to fifteen objectives could be considered with this method. For interpretability, our proposed methodology consists of six steps that consider three viewpoints: objectives, solutions, and variables (i.e., features).To achieve the research goal, we follow a structured approach: first, an extensive literature review that establishes the state-of-the-art and identifies open challenges. Next, empirical analyses of single-objective filter and wrapper methods, as well as multi-objective wrapper methods, are conducted to assess their strengths and limitations. Our MOFS framework is then proposed and evaluated through multiple experiments, including its application to fairness in ML. Finally, the interpretability methodology is instantiated as an interactive dashboard, which is validated through an experimental study involving 50 participants, with statistical analysis to assess its effectiveness.The findings highlight that no single FS method is universally optimal; instead, the best approach depends on dataset characteristics, task requirements, and objectives. While filter methods are computationally efficient and wrapper methods enhance model performance in single-objective settings, the proposed MOFS framework successfully balances multiple conflicting indicators related to performance, complexity, and fairness. Moreover, the interpretability methodology proved essential in helping data scientists to better understand MOFS results, enabling informed decision-making in FS.
  

• RIVERA VIDAL, AMANDA CATALINA: EARTH & FIBERS, Local Technologies Earthen Composites (LoTEC). Local reuse of by-products for zero-waste building materials, appropriate building solutions in vernacular and contemporary architecture.
  
  Author: RIVERA VIDAL, AMANDA CATALINA
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: DOCTORAL DEGREE IN ARCHITECTURAL, BUILDING CONSTRUCTION AND URBANISM TECHNOLOGY
  Department: Department of Architectural Technology (TA)
  Mode: Change of supervisor
  Deposit date: 16/01/2026
  Reading date: 03/02/2026
  Reading time: 15:00
  Reading place: Università Degli Studi di Cagliari
  Thesis director: NAVARRO EZQUERRA, MARIA ANTONIA | ACHENZA, MARIA MADDALENA
  Thesis abstract: This doctoral research addresses the environmental and cultural challenges of the construction sector by re-examining local materials and traditional knowledge, through scientific experimentation. Entitled EARTH & FIBERS Local Technologies Earthen Composites (LoTEC), the thesis investigates earth and vegetal-fiber composites derived from agricultural and excavation by-products as low-impact, circular, and regenerative materials for the decarbonization of the built environment.The study integrates comparative analyses of vernacular construction, experimental material testing, and architectural applications. Case studies from Europe and Latin America document the hygrothermal logic and cultural intelligence of traditional earth-and-fiber systems. Laboratory analyses characterize local raw materials (soils and vegetal residues), leading to the formulation and testing of new earthen composites. Their physical, chemical, and hygrothermal properties—including thermal conductivity, vapor permeability, acoustic absorption, and fire response—have been systematically evaluated. Medium-scale prototypes constructed and monitored under real conditions confirm their technical feasibility and architectural applicability.Results indicate that LoTEC composites display versatile hygrothermal behavior, ranging from thermal insulation to heat storage depending on mixture density and fiber composition. Produced without chemical alteration of the raw materials —only sieved soils and cut vegetal fibers—they require minimal energy for transformation from raw matter to building element. At the end of their service life, they can be rehydrated and reused without performance loss or safely reintegrated into the environment. Their permeability and moisture-buffering capacity make them particularly compatible with the vapor-open nature of historical materials, enabling effective retrofitting and improved indoor comfort while preserving architectural heritage. They also perform efficiently in new constructions based on sufficiency, circularity, and locality.Beyond technical results, the research reframes architectural sustainability within a post-growth paradigm, shifting the focus from expansion and consumption toward care, adaptation, and material responsibility. By merging traditional knowledge with scientific analysis and contemporary design, EARTH & FIBERS advances a model of architecture that minimizes extractive dependence and supports ecological and cultural regeneration, positioning earthen and vegetal-fiber materials as essential components of a decarbonized and contextually grounded built environment.
  

Reading date: 05/02/2026

• TYULNEV, IGOR: Investigation and Control of Phase Transitions by Ultrafast Strong-field Techniques
  
  Author: TYULNEV, IGOR
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: DOCTORAL DEGREE IN PHOTONICS
  Department: Institute of Photonic Sciences (ICFO)
  Mode: Normal
  Deposit date: 10/12/2025
  Reading date: 05/02/2026
  Reading time: 10:00
  Reading place: ICFO Auditorium
  Thesis director: BIEGERT, JENS
  Thesis abstract: This work presents the experiments and results on the application of mid-infrared laser sources towards condensed matter systems for the study and control of manybody interactions within material phases and at phase boundaries. Utilizing the decades in know-how and development of intense, few-cycle waveforms at high repetition rates, the here demonstrated applications leverage the mid-infrared wavelengths to study and control strong-field phenomena at ultrafast time-scales and across phase transitions. To this end non-linear techniques are employed to extend the source capabilities towards a variety of driving and probing wavelengths, meanwhile tailoring spin-angular momentum multi-color beams as driving fields with unique patterns. With strong-field driven dynamics happening at sub-cycle time scales, techniques such as high harmonic generation (HHG) are applied to a variety of materials which undergo electronic and structural transitions. For bulk transition metal dichalcogenides, as the investigated MoS2, the induced spatial and temporal symmetry breaking from a tailored trefoil-shaped strong-field allowed the detection of valley polarization, i.e. a carrier population imbalance between neighboring bandgap extrema. The specific control of the energy bands at these sites, first, allows the realization of a valley switch to be used for optical computing, and second, realizes a hybrid system of light and matter with band topology akin to the Haldane model, which paves the way towards field-induced and controlled topological phase transitions in two-dimensional materials. Furthermore, the field-induced currents and the emerging harmonics are used to probe the potential landscape of the lattice and therefore, simultaneously detect signatures of the crystal and band structure encoded in a static spectrum. Interference within the spectra further reveal the underlying electron-hole dynamics and timings. In high-temperature superconducting ceramics like YBCO, the temperature induced changes in electronic properties are also sensitively detected via HHG, even for more elusive material phases. Meanwhile higher order transitions like the correlated charge density wave (CDW) phase shows a mixture of electronic and structural changes in the HHG crystallography as investigated in TiSe2. The macroscopic and nonlinear approach yields major changes in the harmonic spectra even from small changes in e.g. atom displacement and identifies phase anisotropies which eluded conventional or microscopic techniques.
  

• ZHAO, RUI: Improving SAT and Pseudo-Boolean Solving Technology
  
  Author: ZHAO, RUI
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: DOCTORAL DEGREE IN ARTIFICIAL INTELLIGENCE
  Department: Department of Computer Science (CS)
  Mode: Normal
  Deposit date: 15/12/2025
  Reading date: 05/02/2026
  Reading time: 11:30
  Reading place: Sales d'Actes de la FIB, Campus Nord UPC, Barcelona
  Thesis director: OLIVERAS LLUNELL, ALBERT
  Thesis abstract: The Boolean satisfiability (SAT) problem has seen remarkable progress, from early DPLL and resolution methods to the modern Conflict-Driven Clause Learning (CDCL) paradigm. Nevertheless, significant challenges remain. Theoretically "simple" yet structurally complex problems, such as the pigeonhole principle, continue to challenge state-of-the-art SAT solvers, revealing inherent limitations in core algorithms like CDCL. Although CDCL-based Pseudo-Boolean (PB) solving extends SAT with 0-1 linear arithmetic constraints—enabling more natural modeling and offering exponential speedups in theory—its added complexity introduces computational bottlenecks in propagation, conflict analysis, and optimization. These challenges underscore the need for deeper algorithmic insights and innovative techniques to advance SAT and PB solver performance. This thesis addresses these gaps by advancing the core algorithms and implementation techniques underlying modern SAT and PB solvers. It is structured in two parts:• Part I: SAT Solving – We analyze the limitations of CDCL through both theoretical and practical lenses. The contributions are: (i) new insights from analyzing multiple conflicts, aimed at identifying opportunities to enhance CDCL or understanding the fundamental reasons for the failure of this particular idea; (ii) an empirical study on the equivalence between CDCL solvers and resolution, examining how solvers reproduce unsatisfiability proofs and how decision heuristics and resolution proofs interact.• Part II: Pseudo-Boolean Solving – We introduce optimizations in unit propagation and conflict analysis. Propagation is accelerated through a carefully engineered hybrid technique, while enhanced conflict analysis produces some stronger constraints for more effective search pruning.Beyond performance gains, this work offers profound insights into Boolean constraint reasoning, bridging theoretical gaps and opening new research avenues in SAT, PB, and beyond.
  

Reading date: 06/02/2026

• CAMPOS TORESANO, CRISTINA: Future implications of Maritime Autonomous Surface Ships (MASS) over seafarers’ jobs and Maritime Education and Training (MET). Spanish case- study.
  
  Author: CAMPOS TORESANO, CRISTINA
  Thesis file: (contact the Doctoral School to confirm you have a valid doctoral degree and to get the link to the thesis)
  Programme: NAUTICAL ENGINEERING, MARINE AND NAVAL RADIOELECTRONICS
  Department: Department of Nautical Sciences and Engineering (CEN)
  Mode: Normal
  Deposit date: 13/01/2026
  Reading date: 06/02/2026
  Reading time: 10:00
  Reading place: Sala de Juntes de la Facultat de Nàutica de Barcelona
  Thesis director: CASTELLS SANABRA, MARCEL·LA | BORÉN ALTÉS, CLARA
  Thesis abstract: The main objective of this PhD aims to review, update and implement the requirements for Maritime Education and Training (MET) related to Maritime Autonomous Surface Ships (MASS). In order to achieve this objective, the projects under development and existing autonomous vessels have been studied, as well as new emerging technologies and, consequently, the new knowledge that future crews will need to acquire. The impact of autonomous navigation has been analysed from different perspectives to determine its direct implications on the jobs and training of the future seafarers. Future job creation needs will generate new skills and knowledge that will need to be incorporated into current courses and degrees. These new skills have been analysed from different perspectives, with the aim of proposing a new subject as a first step toward completing the training of future seafarers.First, this thesis conducts a bibliometric analysis to explore the progress and research carried out up to the start of the project, in order to identify the most relevant academic publications and studies in the field of maritime training and autonomous vessels.Secondly, through the use of surveys, the competencies, knowledge, and skills that could be added or adapted are explored, comparing the results obtained with recent advances in education and training. As a result, a proposal is presented for a possible new subject that could constitute a future line of development in the field of autonomous vessel training.Finally, based on the results obtained, a model course has been designed and implemented in collaboration with four European Universities. This course includes an evaluation and analysis of the outcomes to develop a proposed introductory course that could be adopted by all Maritime Education and Training Institutions (METIS).