Candidate: Martin Joseph Aubard

Date, time and location:
25 July 2025, 14:00, Sala de Atos DEEC – I-105, Faculty of Engineering, University of Porto

President of the Jury:
Pedro Nuno Ferreira da Rosa da Cruz Diniz (PhD), Full Professor, Department of Informatics Engineering, Faculty of Engineering, University of Porto

Members:
Bilal Wehbe, Senior Researcher at the German Research Center for Artificial Intelligence, Germany;
Catarina Helena Branco Simões da Silva, Associate Professor, Department of Computer Engineering, Faculty of Science and Technology, University of Coimbra;
Andry Maykol Gomes Pinto, Associate Professor, Department of Electrical and Computer Engineering, Faculty of Engineering, University of Porto;
Ana Maria Dias Madureira Pereira, Coordinating Professor with Aggregation, Department of Computer Engineering, Instituto Superior de Engenharia do Porto, Polytechnic of Porto (Supervisor).

The thesis was co-supervised by Luís Filipe Pinto de Almeida Teixeira (PhD), Associate Professor in the Department of Informatics Engineering at the Faculty of Engineering of the University of Porto.

Abstract:

This thesis addresses the challenge of improving Autonomous Underwater Vehicles (AUVs) onboard detection and interaction capabilities using Side-Scan Sonar (SSS) data. Traditionally, underwater missions relied on pre-defined plans where data are analyzed post-mission by operators or experts. This workflow is time-consuming, often requiring multiple missions to identify and localize underwater targets. The need for repeated missions increases operational costs and complexity, highlighting the inefficiency of current methodologies. Moreover, such approaches do not allow the AUV to interact with detected targets in real time, limiting the scope of mission adaptation and real-time decision-making. To overcome these limitations, this thesis presents a novel framework integrating deep learning models for object detection directly onboard AUVs. This integration enables the vehicle to detect, localize, and interact with underwater targets in real time, offering significant improvements over traditional post-mission analysis. The framework builds upon the LSTS toolchain, which is responsible for AUV motion control and communication, and introduces enhanced real-time data processing capabilities. However, implementing such a model into an embedded system suffers from computational limitations affecting the model’s performance. Thus, the knowledge distillation methods have been implemented, ensuring smaller, more efficient models to perform onboard detection without sacrificing accuracy. Additionally, to improve the model’s robustness against underwater noise, a novel adversarial retraining framework, ROSAR, is introduced, ensuring reliable operation even in noisy sonar environments. Following the onboard detection and localization enhancement, we focused on onboard interaction with the detected object. This is realized by extending the previous onboard framework and validating it through a customized simulator, enhancing interaction with the detected objects, and validating through a pipeline inspection use case, which reduces mission time by combining sonar detection and camera data collection in a single mission, utilizing behavior trees and safety-assessed models. Given the lack of open-source sonar datasets in the field, this thesis contributes to two novel publicly available side-scan sonar datasets, SWDD and Subpipe, which include field-collected data on walls and pipelines and are manually annotated for object detection. By shifting from post-mission analysis to real-time detection and interaction, this thesis significantly improves the operational efficiency of AUV missions. The proposed framework streamlines underwater operations and enhances AUVs’ autonomous behavior, relying on efficient, accurate, and robust object detection model for efficient underwater exploration and monitoring applications.

Candidate:
Filipa Marília Monteiro Ramos Ferreira

Date, time and location:
23 July 2025, 14:30, Sala de Atos, Faculty of Engineering, University of Porto

President of the Jury:
Carlos Miguel Ferraz Baquero-Moreno (PhD), Full Professor, Department of Informatics Engineering, Faculty of Engineering, University of Porto

Members:
Tiago Manuel Lourenço Azevedo (PhD), Associate Researcher, Department of Computer Science and Technology, University of Cambridge, United Kingdom;
Marco António Morais Veloso (PhD), Coordinating Professor, Department of Science and Technology, Oliveira do Hospital School of Technology and Management, Polytechnic Institute of Coimbra;
Luís Filipe Pinto de Almeida Teixeira (PhD), Associate Professor, Department of Informatics Engineering, Faculty of Engineering, University of Porto;
Rosaldo José Fernandes Rossetti (PhD), Full Professor, Department of Informatics Engineering, Faculty of Engineering, University of Porto (Supervisor).

Abstract:

Ensuring the reliability and robustness of deep learning remains a pressing challenge, particularly as neural networks gain traction in safety-critical applications. While extensive research has focused on improving accuracy across datasets, generalisation, interpretability and robustness in the deployment domain remain poorly understood. In fact, in real-world scenarios, models often underperform without clear explanations. Addressing these concerns, uncertainty quantification has emerged as a key research direction, offering deeper insight into neural networks and enhancing confidence, interpretability, and robustness. Among critical applications, self-driving vehicles stand out, where uncertainty-aware object detection can significantly improve perception and decision-making. This thesis explores interpretations of uncertainty tailored to object detection in the context of self-driving vehicles. In this sense, two novel methods to estimate the aleatoric component and one approach to modelling the epistemic uncertainty are proposed. Through the utilisation of anchor distributions readily available in any anchor-based object detector, uncertainty is estimated holistically while avoiding costly sampling procedures. Further, the concept of existence is introduced, a probability measure that indicates whether an object truly exists in the real-world, regardless of classification. Building upon these ideas, three applications of uncertainty and existence are explored, namely the Existence Map, the Uncertainty Map and the Existence Probability. Whilst the aforementioned maps encode the existence measure and the aleatoric uncertainty over the space of input samples, the Existence Probability merges the information provided by the Existence Map with the standard detections, supplementing model outputs. Evaluation showcases the coherence of uncertainty estimates and demonstrates the usefulness of the Existence and Uncertainty Map in supporting the standard model, providing open-set capabilities and giving a degree of confidence to true positives, false positives and false negatives. The merging strategy of the Existence Probability reports a considerable improvement in the performance of the object detector both in validation and perturbation, while detecting all classes of the dataset despite being trained only on cars, pedestrians and cyclists. The second part of this thesis features a study of the underspecification distribution and its connection with the epistemic uncertainty. Underspecification, recently coined, greatly endangers deep learning deployment in safety-critical systems as it depicts the variability of predictors generated by a single architecture with increasingly diverging performance in the application domain. The analysis performed showcases that, if the uncertainty estimates are correctly calibrated, a single predictor is sufficient to predict the spread of the underspecification distribution, avoiding running repeated costly training sessions. All proposed methods are designed to be model-agnostic, real-time compatible, and seamlessly applicable to deployed models without requiring retraining, underscoring their significance for robust and interpretable object detection in autonomous driving.

Candidate:
Lázaro Gabriel Barros da Costa

Date, Time and Location:
18th of July 2025, 16:00, Sala de Atos of the Faculty of Engineering of University of Porto.

President of the Jury:
Pedro Nuno Ferreira da Rosa da Cruz Diniz (PhD), Full Professor, Department of Informatics Engineering, Faculty of Engineering, University of Porto.

Members:
Tanu Malik (PhD), Associate Professor, Department of Electrical Engineering and Computer Science, University of Missouri, U.S.A;
Miguel Carlos Pacheco Afonso Goulão (PhD), Associate Professor, Department of Computer Science, Faculty of Science and Technology, New University of Lisbon;
Gabriel de Sousa Torcato David (PhD), Associate Professor, Department of Informatics Engineering, Faculty of Engineering, University of Porto;
Jácome Miguel Costa da Cunha (PhD), Associate Professor, Department of Informatics Engineering, Faculty of Engineering, University of Porto (Supervisor).

The thesis was co-supervised by Susana Alexandra Tavares Meneses Barbosa (PhD), Senior Researcher at INESCTEC Porto.

Abstract:

Scientific reproducibility and replicability are essential pillars of credible research, especially as computational experiments become increasingly prevalent across diverse scientific disciplines such as chemistry, climate science, and biology. Despite strong advocacy for Open Science and adherence to FAIR (Findable, Accessible, Interoperable, and Reusable) principles, achieving true reproducibility remains a formidable challenge for many researchers. Key issues such as complex dependency management, inadequate metadata, and the often cumbersome access to necessary code and data severely hamper reproducibility efforts. Moreover, existing reproducibility tools frequently offer piecemeal solutions that fail to address the multifaceted needs of diverse and complex experimental setups, particularly those that span multiple programming languages and involve intricate data systems. This thesis addresses these challenges by presenting a comprehensive framework designed to enhance computational reproducibility across a variety of scientific fields. Our approach involved a detailed systematic review of existing reproducibility tools to identify prevailing gaps and limitations in their design and functionality. This review underscored the fragmented nature of these tools, each supporting only aspects of the reproducibility process but none providing a holistic solution, particularly for experiments that require robust data handling or support for many programming languages.
To bridge these gaps, we introduced SCIREP, an innovative framework that automates essential aspects of the reproducibility workflow such as dependency management, containerization, and cross platform compatibility. This framework was rigorously evaluated using a curated dataset of computational experiments, achieving a reproducibility success rate of 94%.
Furthering the accessibility and usability of reproducible research, we developed SCICONV, a conversational interface that simplifies the configuration and execution of computational experiments using natural language processing. This interface significantly reduces the technical barriers traditionally associated with setting up reproducible studies, allowing researchers to interact with the system through simple, guided conversations. Evaluation results indicated that SCICONV successfully reproduced 83% of the experiments in our curated dataset with minimal user input, highlighting its potential to make reproducible research more accessible to a broader range of researchers.
Moreover, recognizing the critical role of user studies in evaluating tools, methodologies, and prototypes, particularly in software engineering and behavioral sciences, this thesis also extends into the realm of experimental tool evaluation. We conducted a thorough analysis of existing tools used for software engineering and behavioral science experiments, identifying and proposing specific features designed to enhance their functionality and ease of use for conducting user studies. These proposed features were validated through a survey involving the research community, confirming their relevance and the need for their integration into existing and future tools. The contributions of this thesis are manifold, encompassing the development of a classification framework for reproducibility tools, the creation of a standardized benchmark dataset for assessing tool efficacy, and the formulation of SCIREP and SCICONV to significantly advance the state-of-the-art in computational reproducibility. Looking forward, the research will focus on expanding the capabilities of reproducibility tools to support more complex scientific workflows, further enhancing user interfaces, and integrating additional functionalities to fully support user studies. By doing so, this work aims to pave the way for a more robust, accessible, and efficient computational reproducibility ecosystem that can meet the evolving needs of the global research community.

Keywords: Reproducibility; Replicability; Reusability; Computational Experiments; Conversational User Interface; User Studies.

Candidate:
Diogo Miguel Filipe Cocharro

Date, time and location:
22nd of July 2025, 15:00, Sala de Atos of the Faculty of Engineering of University of Porto.

President of the Jury:
António Fernando Vasconcelos Cunha Castro Coelho (PhD), Associate Professor in the Department of Informatics Engineering at the Faculty of Engineering of the University of Porto.

Members:
Matt Chiu (PhD), Assistant Professor of Music Theory at the Conservatory of Performing Arts at the Baldwin Wallace University, EUA;
Daniel Gómez-Marín (PhD), Profesor del Departamento de Diseño e Innovación de la Escuela de Tecnología, Diseño e Innovación de la Facultad Barberi de Ingeniería, Diseño y Ciencias Aplicadas de la Universidad Icesi, Colombia;
Sofia Carmen Faria Maia Cavaco (PhD), Assistant Professor in the Department of Computer Science at the Faculty of Science and Technology of Universidade Nova de Lisboa;
Sérgio Reis Cunha (PhD), Assistant Professor in the Department of Electrical and Computer Engineering at the Faculty of Engineering of the University of Porto;
Gilberto Bernardes de Almeida (PhD), Assistant Professor in the Department of Informatics Engineering at the Faculty of Engineering of the University of Porto (Supervisor).

The thesis was co-supervised by Rui Luis Nogueira Penha (PhD), Coordinating Professor of ESMAE – School of Music and Performing Arts.

Abstract:

Music is inherently a temporal manifestation, and rhythm is a crucial component. While rhythm can exist without melody or harmony, the latter cannot exist without rhythm. However, rhythm is often understudied compared to harmony. Rhythmic affinity is a musical concept that describes the natural and pleasing relationship between two or more rhythmic patterns. This happens when these patterns, no matter how complex or seemingly unrelated, come together to create a sense of cohesion and flow rather than dissonance or conflict.
This affinity can arise from various factors, such as shared rhythmic motives, complementary and interlocking rhythmic structures, or a strong underlying pulse that unifies the different layers. For example, two complementary patterns that completely occupy the set of pulses in a cycle by filling each other’s silent pulses with their own active pulses are called interlocking rhythms. These interlocking rhythms are not limited to just the complementary nature of rhythms; we believe they can also be observed in patterns that feature coincident onsets or different underlying pulse grids. This diversity in rhythmic structures represents some of the musical properties we aim to explore in this study.
Music scholars have recently begun to explore affinity-related musical phenomena, particularly building on Harald Krebs’s seminal work on rhythmic dissonance, which offers a comprehensive framework for understanding and categorizing metric dissonance within music. Similarly, Godfried Toussaint’s research examines various methods for measuring rhythmic similarity and for analyzing and generating complementary and interlocking rhythms, providing insights into the structural interrelationships between different rhythmic patterns. Additionally, Clarence Barlow’s work on metrical affinities—often overlooked—contributes important perspectives on the relational characteristics between different meters.
We conducted preliminary experiments to assess the behavior of typical rhythmic similarity metrics across genres. Key findings revealed that similarity varies within a limited range across genres and instruments, which we identify as affinity space. This systematic analysis motivates the discussion and research on the concept of rhythmic affinity, emphasizing the need to understand it as a distinct concept from rhythmic similarity. Furthermore, we identified several limitations that draw this thesis’s main objectives and methodologies, namely the lack of metrics for multi-meter corpus analysis in the context of rhythmic cycles, e.g., loops.
In this context, this study focuses on preprocessing multi-meter representations of rhythmic patterns in the time domain specifically designed for projection in the Discrete Fourier Transform (DFT) space with the goal of exploring rhythmic affinities. We aimed to study the DFT of rhythmic loops towards a mathematical space that reflects metrical levels of alignment (or misalignment), which closely relates to Krebs definition of metric dissonance. This phenomenon relates to practices commonly found in musical composition, such as poly-meter and poly-rhythms, which enable the superimposing of rhythmic patterns that, in principle, show low similarity between each other but that are perceptually pleasing as a combined dissonance, the most known example is the hemiola of a three against two.
Our research follows and extends the body of music theory literature on applying the DFT of pitch classes to distances that reflect human perception and music-theoretical principles. Its application to rhythmic structures is currently limited to particular contexts of a musical piece, not encompassing strategies for multi-meter rhythmic analysis. The main contribution lies in a methodology for multi-meter analysis in the DFT space. Our findings demonstrated that up-sampling the grid of pivotal metrical levels underlying rhythmic pattern representations enables the simultaneous depiction of meters with simple and compound subdivisions. This approach highlights structural relationships within the DFT space, reflected by close distances between related simple and compound metrical templates—for instance, between $4/4$ and $12/8$ or $3/4$ and $6/8$. We implemented this methodology in a prototype system capable of generating rhythmic patterns based on metrical templates and sorting them according to their similarity to a user-defined pattern.

Candidate:
Inês Dias Koch

Date, Time and Location
1st of July 2025, 14:30, Sala de Atos da Faculdade de Engenharia da Universidade do Porto

Title:
“Integration of models for linked data in cultural heritage and contributions to the FAIR principles”

President of the Jury:
João Carlos Pascoal Faria (PhD), Full Professor, Department of Informatics Engineering, Faculdade de Engenharia da Universidade do Porto.

Members:
Maja Žumer (PhD), Full Professor, Department of Library and Information Science of the University of Ljubljana, Slovenia;

María Poveda Villalón (PhD), Associate Professor, Departament of Artificial Intelligence of the Technical University of Madrid, Spain;

José Luís Brinquete Borbinha (PhD), Full Professor, Department of Computer Science and Engineering, Instituto Superior Técnico da Universidade de Lisboa;

Pedro Manuel Rangel Santos Henriques (PhD), Full Professor, Department of Informatics, Escola de Engenharia da Universidade do Minho;

Carla Alexandra Teixeira Lopes (PhD), Associate Professor, Department of Informatics Engineering, Faculdade de Engenharia da Universidade do Porto (Supervisor);

Mariana Curado Malta (PhD), Assistant Professor, Department of Informatics Engineering, Faculdade de Engenharia da Universidade do Porto.

The thesis was co-supervised by Maria Cristina de Carvalho Alves Ribeiro (PhD), Retired Associate Professor in the Department of Informatics Engineering, Faculdade de Engenharia da Universidade do Porto.

Abstract:

The various areas of Cultural Heritage, such as archives, museums, and libraries, have invested in tech-
nological development and innovation to make their resources available to users more efficiently and completely. To this end, the description of these resources is essential so that they are explained in
terms of their context and content, as well as to facilitate their intelligibility and accessibility. In this
sense, each area has begun to develop its own models and standards for describing the cultural objects
it deals with. This has made these standards sector-specific and only able to fulfil the information needs
within the area of knowledge they were developed, exploring only the information described within their
domain. As a result, linking resources from different information sources is challenging.
With the need to make the standards and models more interoperable, linked data models emerged in
Cultural Heritage. These models make it possible to link the various concepts from the different heritage
areas efficiently and effectively, considering the Semantic Web’s characteristics.
In Portugal, the Portuguese National Archives felt the need to develop a linked data model to describe
their cultural objects, which led to the creation of the EPISA Project, the project from which this research
emerged. Thus, this work aims to develop a linked data model to describe archival records, as well as to
connect them with other heritage domains, integrating them with existing linked data models, promoting
the access and reuse of data from heritage institutions based on the specialised description associated
with the cultural objects of these institutions. Additionally, it aims to link existing data models to data from other sources available on the Web, such as Wikidata and DBpedia.
We carry out a study that includes existing data models in Cultural Heritage, such as CIDOC CRM
in museums, RiC-CM in archives, and LRMoo in libraries, along with models that have emerged within
Web projects, such as DBpedia and Wikidata. By describing archival objects, as well as creating and
exploring relationships between other data models, this study identifies common characteristics and
principles, as well as the distinctive aspects of each area. Furthermore, it identifies the possibility of linking elements of the various models, ensuring that the models can be adapted to applications without losing the richness of the conceptualisation carried out in each of the domains.
In a context in which the Web promotes the explicitness of data semantics through the Semantic
Web and provides tools to represent it, it is necessary, on the one hand, to create links between models from different communities and, on the other, to adjust the complexity of each model to each application according to its specific requirements. The FAIR Principles (Findable, Accessible, Interoperable, Reusable) were therefore used as one of the sources for the requirements that data and metadata must fulfil to have a modular structure. We bring together a collection of use cases linked to archives users, including profiles ranging fromcollection managers to heritage promoters and informal users. In addition, we compile and evaluate a set of data modelling experiences using different models.
This work resulted in ArchOnto, a modular ontology that describes archive records. It was developed
considering existing archive standards and validated by experts in the field, specifically archivists from the Portuguese National Archives. ArchOnto is based on CIDOC CRM, combined with four other
specific ontologies also developed in this work.
The development of ArchOnto led to the creation of a prototype platform designed to explore and
manipulate archive records. Additionally, it offers the potential to apply this ontology to other domains, specifically to the representation of cinematographic records.

Keywords: Cultural Heritage; Linked Open Data; Data Integration; Semantic Web; FAIR Principles; Digital Humanities.