Search

Treballs Finals de Grau d'Enginyeria Informàtica, Facultat de Matemàtiques, Universitat de Barcelona, Any: 2021, Director: Laura Igual Muñoz, [en] Atherosclerosis is one of the main precursors to cardiovascular pathologies, the first defunction cause on developed countries. One of its principal diagnosis methodologies is carotid ultrasound images due to their low cost and intrusivity. Nonetheless, these produce low quality representations, which makes the diagnosis of atherosclerotic plaques a laborious task. In spite of that, other risk measurement methodologies exist. Risk tables which, taking into consideration diverse lifestyle and medical data, assign the probability of an individual to suffer a cardiovascular event. These types of tables inherit their functionality from the Framingham study, which analyzed data of United States population to create its risk function, thus being the first study to do so. However, adapting these tables to all population is not precise, as there are different epidemiological factors that can affect the values of the tables, and conducting studies to adjust them is expensive. Moreover, other limitations exist, as it has been proved that most of the future cardiovascular events end up classified on mid-range risk groups, thus not being medicated, besides an age limit to apply the tables, and not accepting missing values. This project sets out to improve the current REGICOR risk function, computed in catalan population, using machine learning prediction models and a combination of medical and ultrasound data of volunteers.

El autor compara los telediarios actuales con el antiguo NO-DO buscando semejanzas y diferencias

El programa desarrollado en el presente Proyecto de Fin de Grado, genera una ciudad a partir de los sistemas denominados L-systems. Es capaz de, definiendo una gramática, crear una red de carreteras que se asemeje a la de las ciudades reales. Una vez creada la red, es posible incluir edifi cios en las parcelas generadas por las diferentes secciones de las carreteras. Para la visualización de la ciudad se utiliza un motor grá fico provisto de las funciones básicas de renderizado: la visualización de objetos 3D, la cámara, texturas, ... Con esas herramientas es posible la visualización de los diferentes estilos de ciudad que se pueden generar mediante el programa.

The following thesis is divided in two main topics. The first part studies variations of optimal prediction problems introduced in Shiryaev, Zhou and Xu (2008) and Du Toit and Peskir (2009) to a randomized terminal-time set up and different families of utility measures. The work presents optimal stopping rules that apply under different criteria, introduces a numerical technique to build approximations of stopping boundaries for fixed terminal time problems and suggest previously reported stopping rules extend to certain generalizations of measures.\ud The second part of the thesis is concerned with analysing optimal wealth allocation techniques within a defaultable financial market similar to Bielecki and Jang (2007). It studies a portfolio optimization problem combining a continuous time jump market and a defaultable security; and presents numerical solutions through the conversion into a Markov Decision Process and characterization of its value function as a unique fixed point to a contracting operator. This work analyses allocation strategies under several families of utilities functions, and highlights significant portfolio selection differences with previously reported results.

Modern Deep Neuronal Network backends allow a great flexibility to define network architectures. This allows for multiple outputs with their specific losses which can make them more suitable for particular goals. In this work we shall explore this possibility for classification networks which will combine the categorical cross-entropy loss, typical of softmax probabilistic outputs, the categorical hinge loss, which extends the hinge loss standard on SVMs, and a novel Fisher loss which seeks to concentrate class members near their centroids while keeping these apart, The authors acknowledge financial support from the European Regional Development Fund and the Spanish State Research Agency of the Ministry of Economy, Industry, and Competitiveness under the project PID2019-106827GB-I00. They also thank the UAM–ADIC Chair for Data Science and Machine Learning and gratefully acknowledge the use of the facilities of Centro de Computaci´on Cient´ıfica (CCC) at UAM