Public display of deposited theses
Presentació d'al·legacions a una tesi doctoral en el termini d'exposició pública
In accordance with the Academic Regulations for Doctoral Studies, doctors may request access to a doctoral thesis in deposit for consultation and, if there are, to send to the Permanent Commission of the Doctoral School the observations and allegations that they consider opportune on the content.
DOCTORAL DEGREE IN ARTIFICIAL INTELLIGENCE
- MONTES GOMEZ, MARIA DE LAS NIEVES: Value Engineering for Autonomous AgentsAuthor: MONTES GOMEZ, MARIA DE LAS NIEVES
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: Article-based thesis
Deposit date: 30/11/2023
Deposit END date: 15/12/2023
Thesis director: SIERRA GARCIA, CARLOS | OSMAN, NARDINE
Committee:
PRESIDENT: CORTÉS GARCÍA, CLAUDIO ULISES
SECRETARI: JONKER, CATHOLIJNTJE MARIA
VOCAL: SEVERINO DE ALMEIDA E PAIVA, ANA MARIA
Thesis abstract: The topic of this thesis is the engineering of values for autonomous agents. This is realised through the formulation, design and implementation of new functionalities for autonomous agents that enable reasoning in terms of values. In particular, we argue for the role of prescriptive norms as value-promoting mechanisms. Hence, value-driven agents should be able to autonomously determine which regulations (such as obligations, permissions or prohibitions) make the Multiagent System they inhabit better promote some values of interest. We lay the foundations of our work on Schwartz¿s Theory of Basic Human Values to establish a consequential connection between values and norms, considering that norms are aligned with respect to values if the outcomes they incentivise satisfy the goals that capture the meaning of values in a particular context. Another feature of Schwartz¿s theory that has been previously overlooked in the literature is the strong social dimension of values. That is, agents should be able to reason not just in terms of their own, but also of the values of others in their community. This points to Theory of Mind (i.e. the cognitive ability to perceive, interpret and reason about others in terms of their mental states) as an outstanding component of value-based reasoning.This thesis is structured around three main contributions (published in journal papers) plus their integration. The first contribution establishes the normvalue relationship as a consequential one in nature, and proposes a methodology for the automated synthesis and analysis of optimally value-aligned normative systems. The second contribution tackles the limitations of the first, and defines the Action Situation Language to systematically express a wide range of rules that may be implemented in a Multiagent System. This language is complemented by a game engine that automatically interprets interaction descriptions and builds their semantics as Extensive Form Games, which are later analysed with standard game-theoretical tools. This leads to a distribution over game outcomes, which are evaluated in terms of their desirability with respect to values. The third contribution introduces Theory of Mind-related functionalities into an existing Belief-Desire-Intention agent architecture, and combines them with abductive reasoning capabilities.The three contributions are integrated in a novel functionality that enables agents to reason about prescriptive norms in terms of dynamic values. This means that an autonomous agent can, at runtime, switch its value perspective to the one it estimates that another agent has. Such perspective-dependent value-based normative reasoning functionality, with its inherent social orientation, constitutes a novel contribution to the community of values for autonomous agents and paves the way for possible applications such as value-based negotiation over normative systems. In summary, value engineering is a principled and systematic approach to computational ethics, which provides an innovative tool set for integrating ethical values into the design of autonomous agents.
- XENOS, ALEXANDROS: Towards a linearly organised embedding space of biological networksAuthor: XENOS, ALEXANDROS
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: 04/12/2023
Deposit END date: 19/12/2023
Thesis director: PRZULJ, NATASA
Committee:
PRESIDENT: BORGWARDT, KARSTEN
SECRETARI: VELLIDO ALCACENA, ALFREDO
VOCAL: HIGHAM, DESMOND JOHN
Thesis abstract: The recent technological advances in high-throughput sequencing have yielded vast amounts of large-scale biological omics data that describe different aspects of cellular functioning. These omics data are typically modelled and analyzed as networks. Due to the high dimensionality of biological networks, embeddings are a cornerstone in the analysis of such complex networks. Embedding biological networks is challenging, as it involves capturing both topological (similar wiring patterns) and neighborhood-based similarity of the nodes. However, current network embedding algorithms do not preserve both types of similarity, which limits the information preserved in the embedding space. Moreover, the existing methods for analyzing the embedding space of molecular networks use the vectors of the biological entities as the input for computationally intensive ML models that aid downstream analysis tasks. In contrast, in the field of NLP, they mine the word embedding space directly by doing simple linear operations between the word embedding vectors. In this thesis, following the NLP paradigm, we mine biological knowledge directly from the embedding space, and we identify the properties of a space that make it suitable for linear operations.In network biology, Non-Negative Matrix Tri Factorization (NMTF) is extensively used to embed networks in a low-dimensional space because it is an explainable AI method that also enables the joint representation of different networks in a shared space. We demonstrate the power of the NMTF-based data integration in the context of COVID-19 by applying two integration frameworks to identify COVID-19 related genes and prioritize drugs for repurposing targeting their gene products. Our newly identified genes could not have been identified with either network-medicine or differential expression based approaches that rely on a single type of omic data.Then, to extract new biological knowledge based on linear operations, we introduce two NLP inspired network representations: the Positive Pointwise Mutual Information (PPMI) matrix and the Graphlet Degree Vector (GDV) PPMI matrix. The PPMI matrix captures the neighborhood-based similarities of the nodes based on random walks between adjacent nodes, while the GDV PPMI matrix, the topological ones by using random walks between similarly wired nodes, independent of being adjacent. As a showcase, we represent the nodes of the human PPI network with our GDV PPMI and PPMI matrix and generate the embedding spaces by factorizing these matrices with NMTF. We show that genes embedded close in these spaces have similar biological functions, so we can extract new biomedical knowledge directly by doing linear operations on their embedded vectors. We exploit this property to predict genes participating in protein complexes and to identify cancer-related genes based on the cosine similarities between the vector representations of the genes.We also go beyond embeddings that preserve one type of similarity by introducing novel random-walk based network embeddings that incorporate the graphlets (small, connected and induced subgraphs) into DeepWalk and LINE methods. We use the graphlets that leverage both topological and neighbourhood-based similarity as the context for the random walks. In a graphlet-based random walk, a node can visit any other nodes that simultaneously participate in the given graphlet. We show that in the graphlet-based representations of the networks, more adjacent nodes have the same label (i.e., the nodes are grouped in a more homophilic way) than in the standard random walk representations. Then we factorize these matrices with NMTF and show that the more homophilic the network representation, the more functionally organized is the corresponding embedding space, and thus the downstream analysis tasks are better. Our new graphlet-based methodologies embed networks in linear spaces, alleviating the need for computationally expensive ML methods.
DOCTORAL DEGREE IN BIOMEDICAL ENGINEERING
- MARIMON SERRA, XAVIER: Caracterització multiescala de dades fisiològiques en la cardiologia translacionalAuthor: MARIMON SERRA, XAVIER
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 BIOMEDICAL ENGINEERING
Department: Department of Automatic Control (ESAII)
Mode: Normal
Deposit date: 01/12/2023
Deposit END date: 18/12/2023
Thesis director: BENITEZ IGLESIAS, RAUL
Committee:
PRESIDENT: GIL MUR, JAVIER
SECRETARI: GOMIS ROMAN, PEDRO
VOCAL: DEMESTRE VILADEVALL, MARIA
Thesis abstract: This thesis aims to explore cardiac cell physiology across various scales, also known as translational cardiology. The research employs novel computational techniques from Biomedical engineering and utilizes medical signal and image databases to investigate the root causes of cardiac contractile system dysfunctions. The study proposes new biomarkers and measures that could aid in diagnosing, prognosticating, and treating cardiac diseases.Besides chapter 1, devoted to the introduction and description of the structure of the thesis, it consists of three main chapters, each of which deals with different stages of cardiac physiology:Chapter 2, focuses on the analysis of currents circulating in ion channels in human atrial cardiomyocytes (HAM), on an intracellular scale. The aim of this chapter is to develop a computational platform for transmembrane ionic currents. This platform allows us to process and extract the kinetic and morphological characteristics of each signal peak that have allowed us to automatically differentiate between healthy and pathological regimes. In conditions where the cardiac cell is stimulated with a time-varying voltage, beat-to-beat electrical alternation phenomena appear, which are indicative of arrhythmogenesis. Based on the characteristics extracted and validated with a synthetic model, a new biomarker has been designed, called the "alternation index", which allows a satisfactory quantification of the degree of alternation between electrical beats.Chapter 3, focuses into the mechanical analysis of single mouse ventricular cardiomyocytes (MVM) at a cellular level, using optical and calcium imaging recordings. The chapter aims to develop a computational video platform using the Digital Image Correlation (DIC) algorithm. This platform calculates displacement fields, strain fields, and sarcolemma length variations during contraction in an isolated cardiomyocyte. By utilizing this technique, potential disruptions in the latently contracting cardiomyocyte during long-term recordings are eliminated, allowing simultaneous monitoring of cardiac contraction and intracellular calcium in a non-invasive and label-free way. The methodology is validated using synthetically created data, and the study applies real experimental data, correlating contractile properties with the calcium signal, Ca2+.Chapter 4 focuses on the analysis of OF1 mouse atrial mechanical contraction signals on a tissue scale. The aim of this chapter is to develop a computational platform that utilizes artificial intelligence (AI) to automatically detect any mechanical contraction irregularities that could cause heart contractile system diseases, such as atrial fibrillation (AF). The study compares traditional AI algorithms based on machine learning (ML) to more advanced ones based on deep learning (DL).This chapter starts with the classification and then the detection of anomalies in mechanical shrinkage signals. First, contraction signals are automatically classified into those without anomalies and those with anomalies, creating a database of labelled contraction records and using supervised learning to determine which AI categorisation algorithms provide the highest accuracy. Secondly, artificial intelligence is used to train an anomaly detector, which determines the precise time at which the anomaly occurs. The various anomalous phenomena observed in the laboratory experiments are modelled in a synthetic database that is used to evaluate and train different anomaly detectors. Finally, the most accurate detector is validated with real experimental signals.
DOCTORAL DEGREE IN NATURAL RESOURCES AND THE ENVIRONMENT
- DELGADO MARTINEZ, AIDA MERCEDES: Desarrollo de una metodología para el diseño de una ruta de turismo sostenible, ¿La Ruta del Oro¿, con base en la valoración de sus patrimonios: geológico, minero, biodiverso, histórico y cultural, primera fase: municipio de El Tambo-Nariño, Colombia. Author: DELGADO MARTINEZ, AIDA MERCEDES
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 NATURAL RESOURCES AND THE ENVIRONMENT
Department: Department of Mining, Industrial and ICT Engineering (EMIT)
Mode: Normal
Deposit date: 04/12/2023
Deposit END date: 19/12/2023
Thesis director: PANTOJA TIMARÁN, FREDDY
Committee:
PRESIDENT: CARRIÓN MERO, PAÚL
SECRETARI: ALFONSO ABELLA, MARIA PURA
VOCAL: BOIXEREU VILA, ESTER
Thesis abstract: The municipality of El Tambo is in the department of Nariño, Colombia, it is an integral part of the Andes Mountain Range, it belongs to the two most important geographical features of this in Colombia, ¿El Nudo de los Pastos¿ and ¿el Macizo Colombiano¿ These regions have the high biodiversity, in the second most megadiverse country in the world. The other hand, One of the most exploited mineral resources in Nariño is gold, through traditional small-scale production processes, mostly located in the Andean zone since pre-Columbian times and, therefore, have a mining and cultural heritage that can be exploited with sustainable methods.The aim of this Thesis was to develop a methodology for the design of a sustainable tourism route, "La Ruta del Oro", through the valuation of the geological, mining, biodiversity, historical and cultural heritage, first phase: Municipality of El Tambo. Its implementation will contribute to generate additional income, especially for women and young people who currently participate indirectly in the mining activity.A systemic and participatory approach was used, where: a) all macro variables are interrelated, b) several disciplines converged and c) the main social representants with knowledge of their territory were involved. The research was developed in 4 phases and 14 steps according to the specific objectives; these phases were: elaboration of the research outline; identification of the key macro-variables and their role; evaluation of the possible locations to integrate "The Gold Route" and proposal of stages and routes for its execution.The main results were:a. An original equation for the Total Valuation of Sites (VTs) based on four criteria and weighting factors, as follows: valuation that the local community gives to the candidate sites, called Preliminary Valuation (VPPs), Valuation of Characteristics and Attributes (VCAs), Valuation of the Visual Quality of the Landscape (VCPs) and Valuation of the Preferences of Potential Visitors from Pasto, capital of Nariño (VPs). The resulting equation is as follows:VTs = 28VPPs+28VCAs+28VCPs+16VPs.b. Fifteen sites were identified and evaluated, corresponding to natural, geological and mining heritage or historical and cultural heritage. With the results obtained, proposals for tours or routes were prepared in three stages. In the first stage, the six sites that obtained the highest scores are linked; among them: Cerro del Espada and Mina la Espada which are part of the thematic axis of the Route. The urban area where the life of the municipality is perceived, in this one is highlighted a) the Basilica where Jesus Nazareno is located b) the local market represents the rural life and c) the Victoriano Salas Arts Hall.c. Relatively new topics in the Latin American scientific context are being discussed in Colombia: the valuation of the landscape, geodiversity, geological heritage and mining heritage.Among the conclusions are:A methodological scheme was designed with a systemic and participatory approach that can be replicated in another territory or in another productive sector. The equation for the total assessment of the candidate sites is its own and original contribution, as is the methodological scheme.El Tambo (N) has some heritage: geological, mining, biodiverse, historical and cultural that deserves to be made known through a Thematic Route called ¿Golden Route.
Last update: 06/12/2023 05:30:20.