Face recognition is a specific case of object recognition. It has received special attention in the recent years due to a great variety of applications such as robot-human interaction, control by gesture, surveillance, security, and people tracking. The idea of face recognition is to give a computer system the ability of finding and recognizing human faces fast and precisely in images or videos. Face recognition aspires to work similar to human perception. Humans identify a big number of known faces, even after years of separation, or under extreme occlusion conditions, e.g. just by looking at a small part of a face. It is a complex task since faces can have different colors, poses, expressions, and sizes or they can be affected by illumination variations or occlusion conditions. Today, there are different methods of face recognition: feature-based approaches (low-level analysis, feature analysis, active shape models) that make explicit use of facial-features such as skin color, facial landmarks or face geometry, and image-based approaches (linear subspace models, neural networks, statistical approaches). The recent increase in the volume of data and computational resources has led to the need for fast and scalable recognition techniques. These techniques should be robust to non-rigid deformations, clutter, occlusion and illumination variations, but, at the same time, they must be sensitive to variations among faces from different persons. For these reasons, neural networks have become a surge of interest Initially, shallow regular neural networks could be used for small image sizes, but they would not scale to deeper networks since a huge amount of parameters would be necessary to be learnt, which can easily cause overfitting. As a result, Convolutional neural networks (CNNs) were proposed. In a CNN, neurons in a layer are connected to small regions of previous layers, which is different to regular neural networks, where all neurons are connected in a full manner. Besides, CNNs make correct assumptions about the nature of the images, for example locality of pixel dependencies \cite{ImageNet}. In general, CNNs have fewer connections and parameters, they are easier to train, and they present similar or better performance than the regular neural networks \cite{ImageNet}. As an extension of the CNNs, deep CNNs were introduced by Imagenet authors because a big number of high-resolution images and powerful GPUs are now available. The authors increased the number of convolutional layers, and they use large receptive fields in the first convolutional layer. As a result, overfitting, which is inherent to the large size of the model, is avoided and effective results are achieved. In addition, the authors of the VGGnet continued adding convolutional layers, but they keep the size of the receptive fields very small (3x3 convolutions) with a stride of 1, throughout the whole network, which decreases the amount of parameters.

Current research on automated vehicles focuses mainly on the drivers of automated vehicles, on its potential to improve the efficiency of traffic operations, safety, congestion and societal benefits, public’s acceptance of automated vehicles as a transport system and the willingness to buy automated vehicles. Nevertheless, there is a research gap in an equally important topic of research; the interactions of the automated vehicles with Vulnerable Road Users (VRU), i.e., cyclists and pedestrians. The WEpods (shuttle buses) are the first automated pods on public roads amidst other traffic, for an extended period of time in the province of Gelderland, in the Netherlands. The main research question revolves around the safety perception of vulnerable road users (VRUs) when interacting with automated vehicles, specifically at unsignalised intersections, and their crossing behaviour in comparison with traditional motor vehicles. The data on road users’ perception was gathered through face-to-face interviews (𝑁=22), a focus group (𝑁=8), and an online survey (𝑁=198). The results of this research showed that in terms of perceived safety, in general, VRUs (pedestrians and cyclists) feel significantly safer when sharing the road with the WEpods (max. speed of 15 𝑘𝑚/ℎ) compared to traditional motor vehicles (max. speed of 30 𝑘𝑚/ℎ). However, cyclists reported feeling less safe when interacting at unsignalised intersections with the automated vehicles, while there was not significant effect on pedestrians. Similarly, pedestrians more often opted for crossing facilities in the presence of the WEpods than in the presence of traditional motor vehicles (this can be interpreted as the result of perhaps feeling less safe), while no significant difference was reported for cyclists. Some of the reasons that could explain the trust in the WEpods are its low operational speed and the trust of most (81.1%) of the VRUs in the automated technology. This makes them expect the WEpods will stop in all possible instances, even when other traffic participants violate traffic rules. Surprisingly, a significant proportion (63.2%) of the participants was not aware of the presence of the steward on board of the WEpods. On the other hand, variables such as the awareness of the steward and having interacted with the WEpods increased the perceived safety amongst VRUs. Moreover, eye contact and gestures use as part of the actual interaction with human drivers of traditional motor vehicles particularly when crossing, was reported to be of importance by the respondents and has also been previously reported in the literature. The VRUs who said that they rely on cues given by drivers, more often indicated a preference to cross at dedicated facilities in the presence of the WEpods than those who stated not to depend on this type of communication. In order to substitute this lack of “real” interaction, information about the WEpods’ operations appeared to be desired by most of the participants in the form of visual information or a mix of auditory and visual. Finally, it was found that individual characteristics of the VRUs, such as their gender and other demographic variables could also have an impact on their perceived safety of VRUs interacting with the WEpods. The findings of the current research appear to point at a prudent attitude of cyclists and pedestrians in their interaction with automated vehicles. Nevertheless, this conservative mindset could be balanced by informing other road users about both, the features (e.g., the presence of a steward on board) and the limitations (e.g., the technology unable to respond to unexpected conditions) of the WEpods. It is considered that this information, along with a suitable communication of intentions of the vehicle to its surroundings are adequate tools to achieve a safe interaction between VRUs and automated vehicles.

Generally studies about identity and consumption have assumed that individuals are immersed in a context of abundance, in which their possibilities to fulfil their needs and desires to purposefully construct a consumption lifestyle are granted. Within this context, studies from the Consumer Culture Theory perspective have described how individuals navigate consumer culture by developing enduring relationships with possessions that help them to build identities and a sense of self. However, this is not always possible during every individual’s life. Certain situational factors not only provoke the individual’s loss of what they have already built, but also lead them to a restricted consumption context. The present study focuses on forced and uncertain changing events, those whose severity and/or rapid change disrupts an individual’s entire life. Through a phenomenological approach of the homeless experience, this study attempts to understand individuals’ transition into a restricted consumption context. The empirical data obtained from a twenty two month ethnographic study evidences that experiencing homelessness, individuals’ loss of possessions and the perceived uncertainty that characterizes life on the streets, trigger not only a self-transformation process, but also alter their relationship with the material world. The findings are contrasted with the current literature extending previous work about transition and consumption by: a) Proposing a new transformational routine to describe the forced self-transformations and how consumption (or the lack of it) becomes an element that influences the deconstruction and construction of a new self. b) Identifying how in a restricted consumption context, individuals develop a different relationship with the material world, and how this changes their modes of consumption.

En esta tesis se aborda el problema de cuánto pescar maximizando el beneficio y disminuyendo el riesgo de colapso mediante una aproximación matemática a tres stocks diferentes, la anchoa en el mar de Alborán, la anchoa en el golfo de Cádiz y la merluza del Atlántico Norte. Teniendo en cuenta las condiciones y particularidades de cada stock, el enfoque para los tres ha sido diferente. Para la anchoa del golfo de Cádiz y del mar de Alborán, aunque se ha utilizado principalmente la interacción entre el medio y el ciclo de vida de la especie, los tipos de forzamiento ambiental y su relación con el ciclo vital son diferentes. Sin embargo, para la merluza se ha tenido en cuenta la relación entre la pesca y su biología, que no se conoce totalmente, ya que el efecto del medio ambiente es menos representativo. En el golfo de Cádiz, ya se conoce la dinámica del forzamiento ambiental causada principalmente por la temperatura, los fuertes vientos y las descargas del río Guadalquivir, y por lo tanto se procede directamente a modelar dichas interacciones con un modelo de doble resolución temporal: Una resolución semanal para las primeras etapas vitales, y mensual para juveniles y adultos. Este modelo logra mejorar estimaciones previas de juveniles y provee estimaciones coherentes de la distribución mensual por tallas de la población mostrando que la doble resolución incorpora adecuadamente la variabilidad del tamaño de la población causada por la influencia del medio ambiente en las etapas tempranas y la dinámica de crecimiento. En el mar de Alborán era necesario analizar primero los factores que podrían estar causando el forzamiento y una indagación exhaustiva logró revelar una conexión entre su sistema circulatorio y el nivel de capturas, donde se ve claramente que el usual alto nivel de energía del sistema circulatorio de la zona no es favorable para el reclutamiento, y que cuando este colapsa es posible la existencia de reclutamientos excepcionales de esta especie. No obstante, la herramienta estadístico-matemática que está presente en todos es el análisis bayesiano que permitió la incorporación de elementos complejos, desde el forzamiento ambiental a la dinámica de capturas y descartes con diferentes flotas, cuantificando la incertidumbre subyacente en todos los procesos involucrados. Las técnicas bayesianas proveen en este trabajo una gran herramienta para asesorar pesquerías con una aproximación más realista a la dinámica biológica y ecológica de las poblaciones. Además, el conocimiento derivado de la aplicación de dichas técnicas se puede proyectar sobre elementos socioeconómicos, diseñando herramientas de gestión que tengan en cuenta el impacto del ecosistema y evaluando su funcionamiento. Por este motivo fue posible proponer en el caso del golfo de Cádiz, una herramienta de evaluación de diversas estrategias para la gestión del recurso, resaltando la importancia de aquellas que tienen en cuenta la interacción entre el ecosistema y la biología de la especie estudiada. La herramienta se basa en la aplicación de un esquema de seguros que permite calcular el valor de la información ambiental aplicada a la gestión, usando conceptos como ingreso medio, prima y riesgo de colapso. Los resultados de una estrategia propuesta que incluye el forzamiento ambiental muestran que el riesgo económico no se disminuye, dado que la volatilidad del ingreso y el valor de la prima son mayores, pero si disminuye el riesgo de colapso y se genera un monto de ingresos tal que permite compensar dicha volatilidad. El aumento considerable de ingresos obtenidos bajo esta estrategia hacen que esta herramienta sugiera la posibilidad de estabilizar ingresos sobre un recurso inherentemente fluctuante. Por tanto, representa un precedente para la posibilidad de un manejo cooperativo del riesgo que responda a una cuota variable determinada por el medio ambiente. A pesar de las múltiples ventajas de las técnicas bayesianas aplicadas tanto al modelado del ecosistema y sus interacciones con la pesca, como a la gestión del recurso, el tiempo computacional es demasiado elevado para modelos con un gran número de parámetros y procesos como el de la merluza. El rápido avance tecnológico resolverá este inconveniente y hará que esto sea posible en el futuro, permitiendo que la medida de incertidumbre que provee esta herramienta pueda ser la base de la toma de decisiones relevantes para el sector pesquero. Resumen completo tomado de esta tesis.

Abstract: Greater local economic activity could improve children’s educational attainment by expanding the resources that could be invested in education. However, increased economic activity could also result in limited educational gains, as children may be induced into employment. This paper studies violence as an alternative explanation for the limited gains, focusing on Colombia, an oil-producing country with a long-standing civil conflict. Existing evidence shows that oil price shocks fuel civil conflicts; I explore whether increased violence undermines any positive effect greater oil resources might have on educational attainment by considering exogenous increases in international oil prices and geographic variations in oil production. Estimates suggest rising oil prices between 1998 and 2005 had limited effects on the number of years of schooling and whether children are behind grade for their age. Moreover, oil price shocks had small but surprising negative effects on primary school enrollment. My analysis reveals that the negative effects on younger children education might be driven by the aforementioned increase in civil conflicts and by additional oil resources being invested in sectors other than education. Resumen: Una economía más dinámica puede mejorar la escolaridad de los niños, a través de los mayores recursos que se inviertan en educación. Sin embargo, un mayor dinamismo también podría resultar en pocas ganancias educativas, si incentiva el trabajo infantil. El presente artículo explora la violencia como una explicación alterna a este fenómeno, enfocándose en Colombia, un país productor de petróleo que ha sufrido de un largo conflicto armado. Con base en evidencia existente que documenta la intensificación del conflicto armado causado por los aumentos en los precios del petróleo, éste artículo investiga si la mayor violencia destruye el efecto positivo de los mayores recursos petroleros en la escolaridad de los niños. Este artículo utiliza el aumento exógeno en los precios del petróleo y la variación geográfica en la producción de petróleo para identificar dicho efecto. Los resultados sugieren que el aumento en los precios del petróleo de 1998 a 2005 no tuvo ningún efecto en los años de escolaridad de los niños, ni en su probabilidad de estar atrasados en el colegio. Más aún, los aumentos en los precios del petróleo tuvieron efectos pequeños pero negativos en la matricula primaria. El análisis sugiere que éstos efectos negativos pueden darse debido a la intensificación del conflicto armado y a la inversión de los recursos petroleros en sectores diferentes a la educación.

En este capítulo se muestran claros indicios de que la disminución en la fuerza del chorro atlántico (corriente de agua con dirección Este, pasando por el Estrecho de Gibraltar) permite una mayor supervivencia de la anchoa en el mar de Alborán, en sus estadios más tempranos, aumentando así las probabilidades de un reclutamiento exitoso y, en consecuencia, un alto número de capturas. Texto tomado y adaptado de la “Introducción general” de esta tesis.

Resumen: Este archivo contiene el resumen de la tesis; la “Introducción general”; los objetivos generales y específicos, diferenciados por capítulos, y el capítulo 1. En tal capítulo se desarrolla un nuevo modelo bayesiano para la anchoa en el golfo de Cádiz. Dicho modelo forma parte de esta tesis y fue aceptado para su publicación en el Canadian Journal of Fisheries and Aquatic Sciences. Es un modelo con resolución temporal semanal para las primeras etapas vitales, y mensual para juveniles y adultos que incorpora la influencia del medio ambiente para etapas tempranas y una estructura de crecimiento que permite estimar la distribución mensual de tallas. Este modelo, además, provee una distribución mensual de las capturas a partir de los parámetros estimados, lo que permite simular series temporales y comparar como se comportarían las capturas usando diferentes estrategias de pesca para diseñar una que sea óptima. Texto tomado y adaptado de la “Introducción general” de esta tesis.

Active portfolio management aims to deliver superior returns through using an extensive analysis of securities in order to identify mispriced stocks and estimate alphas. Moreover, an active strategy relies on the capability of the active manager to transform his/her forecasting skills into an active portfolio. This thesis assesses the effects of investment constraints in the ex ante capabilities of an active manager to construct portfolios through the fundamental law of active management, and provides evidence on the efficiency loss of the manager in presence of such constraints through a Monte Carlo simulation. 

Based on the characteristics of the study area, and those of the fauna studied, corridor designs were proposed using a methodology that considered three main elements: 1) Evaluation of habitat use by wildlife; 2) Use of a Geographical Information System (GIS) to define corridor boundaries (pathways between habitat quality ranks); 3) Management regulations using the corridor as part of the landscape. Herbivores were consistently present in those areas covered by primary forest. Omnivores were located in a wider variety of covers (grass and agriculture, secondary forest, primary forest, melina). Wildcats were located in primary and secondary forest cover types (mainly used by their herbivore prey). The three guilds were maintained far from highways and human settlements but near forest and water resources. Several species crossed the highways that surround the Golfo Dulce Forest Reserve (GDFR). Maps show that available habitat areas of ""excellent"" and ""good"" habitat quality features, were reduced for the species Panthera onca and Puma concolor, and their conflict areas were more extensive; this fact probably pushed these wildcats to move into heavily altered environments. This study reinforced affirmations about the necessity of corridors, based on identification of habitat that meets minimum quality requirements for wildlife (corridor users), and that defines or modifies their spatial distribution. The pathways designed by this procedure coincide with prior studies showing that critical areas for all wildcats species (and for other species related to them) were located north of Corcovado National Park (CNP), San Juan, Mogos, and the area where Piedras Blancas National Park (PBNP) borders the Golfo Dulce Forest Reserve (GDFR). However, the spatial location of corridors is quite different in this proposal and other possible pathway was identified for P. onca and its prey. This pathway was found to be in conflict toward Drake and wetlands of Sierpe Terraba too.

Various town centres in the UK have been experiencing worrying symptoms, such as store closures, reduction of footfall, and decrease of retail floor-space. Specifically for the town centres in Greater Manchester, none of the alternatives considered for addressing this problem contemplated the potential of smartphones, personalisation, and social networks. This dissertation is motivated by the current relevance of these topics in different contexts; businesses, social and individual behaviour, retailers, and so forth. The goal of this dissertation is to develop a conceptual model where town centres use technologies and information about individuals to influence their attitude to visit the town centre. Explicitly, the model uses digital footprints in social media and social networks, and data from mobile and static sensors, to build a profile of UK smartphone users. From the information inferred, a personalised message or service can be delivered to the smartphone user which would influence his or her attitude towards visiting the town centre. The model is based on information flowing in both directions benefiting the user as well as the town centre. As sharing, extracting, and using information is an important part of the conceptual model, it is essential that it transmits trust to the users about the privacy of the data, and about the way it is handled. However, the empirical test of the model was not within the scope of this dissertation. Hence, it is proposed to be undertaken as a further research.