Search

El Learning Dashboard és una eina de visualització i monitoratge d'assoliment d'objectius d'aprenentatge per equips d'estudiants que realitzen projectes software. Aquest recull i analitza dades de diverses fonts per calcular i visualitzar mètriques, factors de qualitat i indicadors estratègics. Va ser implementat com a una adaptació del projecte Q-Rapids dashboard. Durant aquesta adaptació, alguns components van deixar de ser funcionals. L'objectiu d'aquest projecte és afegir al Learning Dashboard dues funcionalitats ja existents en el Q-Rapids Dashboard que per el moment no han pogut ser adaptades amb èxit. La primera és un component d'alertes, que envia alertes quan uns valors determinats per a les mètriques, els factors i els indicadors estratègics de seguiment són traspassats. La segona funcionalitat és un component de prediccions, que preveu l'evolució de les diferents mètriques, factors i indicadors estratègics., The Learning Dashboard is a tool for visualizing and monitoring the achievement of learning objectives in student's teams developing software projects. The Learning Dashboard gathers and analyzes information from several sources to calculate and visualize metrics, quality factors and strategic indicators. It was implemented as an adaptation of the Q-Rapids dashboard project. During this adaptation/integration, some components stopped being functional. The objective of this project is to implement in the Learning Dashboard two functionalities already existing in the Q-rapids Dashboard that haven't been adapted successfully. The first one is an alert component, which sends alerts when certain values ¿¿for the tracking metrics, factors and strategic indicators are breached. The second one, a prediction component which predicts the evolution of the different indicators.

El aumento del consumo energético en los últimos años ha provocado un avance tecnológico que ha hecho surgir varios problemas con los combustibles fósiles, los cuales son uno de los principales causantes del calentamiento global. Para conseguir reducir las emisiones, se debe buscar una forma de generación de energía limpia. Dentro del grupo de energías con bajas emisiones, la energía nuclear es una opción por la que han optado algunos países, al tratarse de una forma de generación sin emisiones y capaz de producir energía eléctrica de forma continua. Por otro lado, este avance tecnológico ha traído consigo también la aparición de la inteligencia artificial (IA). Esta tecnología ha llegado a nuestras vidas con el objetivo de facilitar el trabajo al ser humano, y se hace uso de ella en una gran diversidad de situaciones. Dentro del campo de la inteligencia artificial se encuentra el aprendizaje profundo o Deep Learning, método mediante el cual las redes neuronales aprenden a distinguir patrones y características significativas de un problema en concreto, teniendo así la capacidad de acelerar alta variedad de procesos. En este trabajo se hará uso de esta tecnología con el objetivo de diseñar una red neuronal que acelere el proceso de cálculo de ciertos parámetros de elementos combustibles irradiados para, en un paso posterior no objeto de este trabajo, poder llegar a optimizar el almacenamiento de elementos combustibles en contenedores con la ayuda de un algoritmo genético. Para ser capaces de conseguir lo propuesto, se necesita un set de datos lo suficientemente grande para entrenar de forma correcta la red neuronal y comprobar su precisión. A fin de conseguir un set de datos de tal tamaño y variedad, se han generado de forma semi-aleatoria características y ciclos de irradiación de elementos combustibles y se ha calculado mediante el programa SCALE, la criticidad y composición de cada uno de ellos tras el enfriamiento, estos valores son los que calculará la red neuro, Azken urteko energia-kontsumoaren gorakadak aurrerapen teknologiko bat eragin du, eta horrek hainbat arazo sortu ditu erregai fosilekin, horiek baitira berotze globalaren eragile nagusietako bat. Emisioak murrizteko energia garbia sortzeko era bat bilatu behar da. Emisio gutxi aske egiten dituzten energien taldearen barruan, herrialde batzuk aukeratu dute energia nuklearra, emisiorik gabeko sorkuntza modu bat eta energia elektrikoa etengabe ekoizteko gai delako. Bestalde, aurrerapen teknologiko horrek inteligentzia artifizialaren agerpena ere ekarri du. Teknologia hau pertsonei lana errazteko helburuarekin iritzi da gure bizitzara, eta hainbat egoeretan erabiltzen da. Inteligentzia artifizialaren zelaian, ikaskuntza sakona edo Deep Learning metodoa dago. Metodo horren bidez, sare neuronalek arazo jakin baten patroiak eta ezaugarri garrantzitsuak bereizten ikasten dute, eta, horrela, prozesu-aniztasun handia bizkortzeko gaitasuna dute. Lan honetan, teknologia hori erabiliko da, elementu erregai irradiatuen parametro jakin batzuk kalkulatzeko prozesua azkartuko duen sare neuronal bat diseinatzeko, ondoren, lan honen helburua ez den urrats batean, elementu erregaiak edukiontzietan biltegiratzea optimizatu ahal izateko algoritmo genetiko baten laguntzarekin. Proposatutakoa lortzeko gai izateko, datu-set handi bat behar da sare neuronala behar bezala entrenatzeko eta zehaztasuna egiaztatzeko. Tamainu eta barietate horretako datu-set bat lortzeko, elementu erregaien ezaugarriak eta irradiazio-zikloak sortu dira erdi ausaz, eta SCALE programaren bidez, hoztu ondoren horietako bakoitzaren kritikotasuna eta konposizioa kalkulatu dira. Balio horiek dira neurona-sareak behin entrenatuta kalkulatuko dituenak. Eskuragarri dagoen datu-setarekin, sarea diseinatzen da, geruza kopurua, geruza bakoitzeko neuronak eta sarearen diseinuari, doikuntzari eta entrenamenduari buruzko beste parametro batzuk aldatuz, iragarpenetan errorea minimizatzeko. Azterketa egin ondoren, aukeratutako sa, Technological development has triggered in an increase of electric consumption. Along with this, some problems related to fossil fuels, which are the one of the main reasons of global warming, have arisen. To be capable of reducing emissions, a clean way of generating electrical energy has to be found. Inside the category of low emission energies, some countries have taken the path of nuclear energy because in addition to having low emissions, it is capable of continuously generating electricity. On the other hand, this technological development has brought about the rise of artificial intelligence. This technology has come to our lives to make them easier, and people use it in many ways. Inside the field of artificial intelligence, there lays Deep Learning, a method where a neural network learns to distinguish significant characteristics and patterns so that it acquires the ability to accelerate a wide range of processes. This work will make use of the technology mentioned above with the objective of designing a neural network which can speed up the calculation process of certain parameters of spent fuel assemblies so that, in a future project out of the scope of this work, it could optimize the used nuclear fuel canister loading with the help of a genetic algorithm. In order to be capable of doing this, a set of data big enough to train and test the neural network will be needed. When it comes to building a wide and varied set of data, the burnup of several fuel assemblies semi-randomly generated has been simulated in SCALE, obtaining the value of the criticality and composition of certain isotopes after an also random cooling time. Those will be the values calculated by the neural network once designed. With a complete set of data, the network is designed varying the number of neurons and layers and training it to achieve the lowest error possible. Once this work is finished, the type of network chosen can be observed and the network will have the capacity of pre

This entry departs from promissory imaginaries about educational technology to show how postdigital research adopts a critical stance on futures. Futures are observed as both a research object and a methodological frame to understand an increasingly complex and technology-dense reality. Temporalities are identified as follows: futures in the present, near futures, far futures, and alternative futures. These situated time-spaces cater for different methodological approaches. Anticipatory methods are elaborated in relation to a near future, while far futures can be grasped through speculatively methods. Alternative futures are usually formulated in the literary vein of science fiction. The tendency in recent futures studies is to make explicit dimensions of ethics, care, and values in the crafting of educational futures.

The growth of populations is of interest in a broad variety of fields, such as epidemiology, economics or biology. Although a large variety of growth models are available in the scientific literature, their application usually requires advanced knowledge of mathematical programming and statistical inference, especially when modelling growth under dynamic environmental conditions. This article presents the biogrowth package for R, which implements functions for modelling the growth of populations. It can predict growth under static or dynamic environments, considering the effect of an arbitrary number of environmental factors. Moreover, it can be used to fit growth models to data gathered under static or dynamic environmental conditions. The package allows the user to fix any model parameter prior to the fit, an approach that can mitigate identifiability issues associated to growth models. The package includes common S3 methods for visualization and statistical analysis (summary of the fit, predictions,…), easing result interpretation. It also includes functions for model comparison/selection. We illustrate the functions in biogrowth using examples from food science and economy.

© Author(s) 2009. Chiara Cagnazzo is supported by the Centro Euro-Mediterraneo per i Cambiamenti Climatici. Elisa Manzini acknowledges the support of the EC SCOUT-O3 Integrated Project (505390-GOCE-CT-2004) for part of this work. Natalia Calvo was supported by the Spanish Ministry of Education and Science and the Fulbright Commission in Spain. CCSRNIES’s research has been supported by the Global Environmental Research Fund (GERF) of the Ministry of the Environment (MOE) of Japan (A-071). MRI simulations have been made partly with the MRI supercomputer and partly with the NIES supercomputer. CMAM simulations were supported by the Canadian Foundation for Climate and Atmospheric Sciences and run on the Environment Canada Supercomputer. We acknowledge the modeling groups for making their simulations available for this analysis, the Chemistry-Climate Model Validation Activity (CCMVal) for WCRP’s (World Climate Research Programme) SPARC (Stratospheric Processes and their Role in Climate) project for organizing and coordinating the model data analysis activity, and the British Atmospheric Data Center (BADC) for collecting and archiving the CCMVal model output. Chiara Cagnazzo and Elisa Manzini are grateful to Antonio Navarra for useful discussions. We are thankful to John Austin for suggestions and discussions on the manuscript., The connection between the El Nino Southern Oscillation (ENSO) and the Northern polar stratosphere has been established from observations and atmospheric modeling. Here a systematic inter-comparison of the sensitivity of the modeled stratosphere to ENSO in Chemistry Climate Models (CCMs) is reported. This work uses results from a number of the CCMs included in the 2006 ozone assessment. In the lower stratosphere, the mean of all model simulations reports a warming of the polar vortex during strong ENSO events in February-March, consistent with but smaller than the estimate from satellite observations and ERA40 reanalysis. The anomalous warming is associated with an anomalous dynamical increase of column ozone north of 70 degrees N that is accompanied by coherent column ozone decrease in the Tropics, in agreement with that deduced from the NIWA column ozone database, implying an increased residual circulation in the mean of all model simulations during ENSO. The spread in the model responses is partly due to the large internal stratospheric variability and it is shown that it crucially depends on the representation of the tropospheric ENSO teleconnection in the models., Canadian Foundation for Climate and Atmospheric Sciences, EC SCOUT-O3 Integrated Project, Spanish Ministry of Education and Science, Ministry of the Environment (MOE) of Japan, Fulbright Commission in Spain, Euro-Mediterraneo per i Cambiamenti Climatici, Depto. de Física de la Tierra y Astrofísica, Fac. de Ciencias Físicas, TRUE, pub

In the past few years, online convex optimization (OCO) has received notable attention in the control literature thanks to its flexible real-time nature and powerful performance guarantees. In this article, we propose new step-size rules and OCO algorithms that simultaneously exploit gradient predictions, function predictions and dynamics, features particularly pertinent to control applications. The proposed algorithms enjoy static and dynamic regret bounds in terms of the dynamics of the reference action sequence, gradient prediction error, and function prediction error, which are generalizations of known regularity measures from the literature. We present results for both convex and strongly convex costs. We validate the performance of the proposed algorithms in a trajectory tracking case study, as well as portfolio optimization using real-world datasets., Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public., Team Peyman Mohajerin Esfahani, Team Tamas Keviczky

Small to medium-sized man-made freshwater reservoir are a reliable source for drinking water supply, hydropower generation and irrigation purposes worldwide. However, water volumes in these reservoirs can be significantly affected by prolonged droughts, resulting in severe impacts on society (Kozacek, 2014; Mahr, 2018). To mitigate the impact of such events it is crucial for decision makers to know when the available water resources are lacking. Although many reservoirs are closely monitored, this data is not always readily available. Inadequate information sharing, inaccessibility, and a lack of tools to predict future reservoir storages contribute to this problem. Remote sensing has the potential to address this problem. The Global Water Watch is a platform that and provides earth-observed surface area dynamics that can be used to monitor small to medium-sized reservoirs worldwide and detect trends in water availability. While this method serves as a valuable indicator of water availability, it falls short in providing decision-makers with the necessary absolute volume time series and volume predictions. Currently, no platform exists beyond in-situ measurements to meet this essential need. This thesis presents a novel method for retrieving near real-time volume time series in small to medium-sized man-made reservoirs worldwide using remotely sensed open data. The method utilises the MERIT-Hydro digital elevation model, HydroMT and stream flow methods by Eilander et al. (2023), and literature by Messager et al. (2016) to reconstruct reservoir bathymetry. This novel approach in reconstructing reservoir bathymetry enables the conversion of available reservoir area time series into volume time series. These were employed in autoregressive and multi-linear regression models to predict water availability up to six months in advance. The models incorporate ERA5 precipitation data by Hersbach's (2020) and the Standardised Precipitation and Evaporation In, Reservoir bathymetry plots, cross-sections, hypsometry for the Indian reservoirs provided in the supplementary materials., Global Water Watch, Water Management | Hydrology

The general feeling is that no predictions can be made based on agent-based social simulations. The outcomes of social simulations are based on the behaviors of individuals and their interactions. Behavioral models are always incomplete and often, also incorrect with respect to real behavior and thus the outcomes of agent-based social simulations cannot be trusted as predictions. In this article, we argue that behavioral models do not have to be valid in all respects, but only in the essential aspects in order to be able to make useful predictions. Based on some case studies on the effectiveness of restrictions during the COVID-19 crisis, we show that what are essential aspects of a behavioral model that need to be valid depends on the specific situation that is simulated. The predictions that were needed for the COVID-19 crisis were made with an agent-based social simulation framework using a behavioral model based on needs. The predictions could indicate the relative increase or decrease of COVID-19 infections due to the introduction of a new restriction. It shows that useful predictions can be made based on social simulations, but that we have to be careful on what type of predictions to make.

Les maladies infectieuses font partie des causes de mortalité et de morbidité les plus importantes au monde avec en moyenne chaque année 9.6 millions de décès humains. Il existe une très grande diversité de pathogènes responsables de ces maladies qui peuvent être naturellement présents dans l’environnement ou être issus de sources de contaminations spécifiques ponctuelles (PS ou Point Sources; p. ex.: contamination fécale directe d’excréments d’animaux ou d’humains, fuites d’eau usée) ou diffuses (NPS ou Non-Point Sources; p. ex.: épandage agricole, eaux résiduelles urbaines). Les pathogènes peuvent aussi circuler entre les différents hôtes (c.à.d.: humains, animaux, végétaux) et vecteurs (p. ex.: rongeurs, moustiques, tiques) qu’ils contaminent. Ainsi jusqu’à 60.3% des maladies infectieuses émergeantes chez l’Humain seraient d’origine animale. Pour limiter l’impact des maladies sur les populations, des connaissances sur la distribution géographique des pathogènes et leurs niches écologiques sont nécessaires. En effet, endiguer de futures éclosions de maladies est un des défis de la science actuelle pour soutenir la santé publique, la sécurité alimentaire et la conservation des espèces. Actuellement, seulement 4% des 355 pathogènes infectieux les plus importants pour l’Humain ont été cartographiés à l’échelle mondiale. Par ailleurs, les changements environnementaux et climatiques ont des conséquences sur l’occurrence des maladies en altérant les niches écologiques des pathogènes, de leurs hôtes et de leurs vecteurs. En ce sens, les connaissances déjà établies pourraient, à l’avenir, devenir obsolètes sans le développement de modèles prédictifs fiables. Dans ce projet, une approche télé-épidémiologique et macroécologique a été utilisée pour développer des modèles prédictifs sur différents types de pathogènes potentiels échantillonnés dans des lacs canadiens. Cette approche se base sur l’idée que les lacs sont des sentinelles des changements environnementaux qui se pr, Infectious diseases are among the most important causes of mortality and morbidity in the world with an average of 9.6 million human deaths per year. There is a wide variety of pathogens responsible for these diseases, which may be naturally present in the environment or may originate from specific point (PS; e.g., direct fecal contamination from animal or human feces, wastewater leaks) or non-point sources (NPS; e.g., agricultural land application, urban wastewater) of contamination. Pathogens can also circulate between the different hosts (i.e., humans, animals, plants) and vectors (e.g., rodents, mosquitoes, ticks) that they contaminate. Thus, up to 60.3% of emerging infectious diseases in humans are of animal origin. To limit the impact of diseases on populations, knowledge of the geographical distribution of pathogens and their ecological niches is necessary. Indeed, containing future disease outbreaks is one of the challenges of current science to support public health, food security and species conservation. Currently, only 4% of the 355 most important pathogens for humans have been mapped on a global scale. In addition, environmental and climatic changes have consequences on the occurrence of diseases by altering the ecological niches of pathogens, their hosts, and vectors. In this sense, the knowledge already established could become obsolete in the future without the development of reliable predictive models. In this project, a tele-epidemiological and macroecological One Health approach was used to develop predictive models on different types of potential pathogens sampled in Canadian lakes. This approach is based on the idea that lakes are sentinels of environmental changes occurring in their watersheds. They could capture both naturally occurring microbes and microbes from localized sources in the watersheds. The multidisciplinary method developed in this project involves the use of microbial data from DNA sequences and bio-optical properties

The general feeling is that no predictions can be made based on agent-based social simulations. The outcomes of social simulations are based on the behaviors of individuals and their interactions. Behavioral models are always incomplete and often, also incorrect with respect to real behavior and thus the outcomes of agent-based social simulations cannot be trusted as predictions. In this article, we argue that behavioral models do not have to be valid in all respects, but only in the essential aspects in order to be able to make useful predictions. Based on some case studies on the effectiveness of restrictions during the COVID-19 crisis, we show that what are essential aspects of a behavioral model that need to be valid depends on the specific situation that is simulated. The predictions that were needed for the COVID-19 crisis were made with an agent-based social simulation framework using a behavioral model based on needs. The predictions could indicate the relative increase or decrease of COVID-19 infections due to the introduction of a new restriction. It shows that useful predictions can be made based on social simulations, but that we have to be careful on what type of predictions to make.

The general feeling is that no predictions can be made based on agent-based social simulations. The outcomes of social simulations are based on the behaviors of individuals and their interactions. Behavioral models are always incomplete and often, also incorrect with respect to real behavior and thus the outcomes of agent-based social simulations cannot be trusted as predictions. In this article, we argue that behavioral models do not have to be valid in all respects, but only in the essential aspects in order to be able to make useful predictions. Based on some case studies on the effectiveness of restrictions during the COVID-19 crisis, we show that what are essential aspects of a behavioral model that need to be valid depends on the specific situation that is simulated. The predictions that were needed for the COVID-19 crisis were made with an agent-based social simulation framework using a behavioral model based on needs. The predictions could indicate the relative increase or decrease of COVID-19 infections due to the introduction of a new restriction. It shows that useful predictions can be made based on social simulations, but that we have to be careful on what type of predictions to make.

The general feeling is that no predictions can be made based on agent-based social simulations. The outcomes of social simulations are based on the behaviors of individuals and their interactions. Behavioral models are always incomplete and often, also incorrect with respect to real behavior and thus the outcomes of agent-based social simulations cannot be trusted as predictions. In this article, we argue that behavioral models do not have to be valid in all respects, but only in the essential aspects in order to be able to make useful predictions. Based on some case studies on the effectiveness of restrictions during the COVID-19 crisis, we show that what are essential aspects of a behavioral model that need to be valid depends on the specific situation that is simulated. The predictions that were needed for the COVID-19 crisis were made with an agent-based social simulation framework using a behavioral model based on needs. The predictions could indicate the relative increase or decrease of COVID-19 infections due to the introduction of a new restriction. It shows that useful predictions can be made based on social simulations, but that we have to be careful on what type of predictions to make.

The general feeling is that no predictions can be made based on agent-based social simulations. The outcomes of social simulations are based on the behaviors of individuals and their interactions. Behavioral models are always incomplete and often, also incorrect with respect to real behavior and thus the outcomes of agent-based social simulations cannot be trusted as predictions. In this article, we argue that behavioral models do not have to be valid in all respects, but only in the essential aspects in order to be able to make useful predictions. Based on some case studies on the effectiveness of restrictions during the COVID-19 crisis, we show that what are essential aspects of a behavioral model that need to be valid depends on the specific situation that is simulated. The predictions that were needed for the COVID-19 crisis were made with an agent-based social simulation framework using a behavioral model based on needs. The predictions could indicate the relative increase or decrease of COVID-19 infections due to the introduction of a new restriction. It shows that useful predictions can be made based on social simulations, but that we have to be careful on what type of predictions to make.

Species Distribution Models (SDMs) are essential tools for predicting climate change impact on species’ distributions and are commonly employed as an informative tool on which to base management and conservation actions. Focusing only on a part of the entire distribution of a species for fitting SDMs is a common approach. Yet, geographically restricting their range can result in considering only a subset of the species’ ecological niche (i.e., niche truncation) which could lead to biased spatial predictions of future climate change effects, particularly if future conditions belong to those parts of the species ecological niche that have been excluded for model fitting. The integration of large-scale distribution data encompassing the whole species range with more regional data can improve future predictions but comes along with challenges owing to the broader scale and/or lower quality usually associated with these data. Here, we compare future predictions obtained from a traditional SDM fitted on a regional dataset (Switzerland) to predictions obtained from data integration methods that combine regional and European datasets for several bird species breeding in Switzerland. Three models were fitted: a traditional SDM based only on regional data and thus not accounting for niche truncation, a data pooling model where the two datasets are merged without considering differences in extent or resolution, and a downscaling hierarchical approach that accounts for differences in extent and resolution. Results show that the traditional model leads to much larger predicted range changes (either positively or negatively) under climate change than both data integration methods. The traditional model also identified different variables as main drivers of species’ distribution compared to data-integration models. Differences between models regarding predicted range changes were larger for species where future conditions were outside the range of conditions existing in the region

Airlines are critical today for carrying people and commodities on time. Any delay in the schedule of these planes can potentially disrupt the business and trade of thousands of employees at any given time. Therefore, precise flight delay prediction is beneficial for the aviation industry and passenger travel. Recent research has focused on using artificial intelligence algorithms to predict the possibility of flight delays. Earlier prediction algorithms were designed for a specific air route or airfield. Many present flight delay prediction algorithms rely on tiny samples and are challenging to understand, allowing almost no room for machine learning implementation. This research study develops a flight delay prediction system by analyzing data from domestic flights inside the United States of America. The proposed models learn about the factors that cause flight delays and cancellations and the link between departure and arrival delays.

During service life, road tunnels may face threats of accidental explosions, e.g., Boiling Liquid Expansion Vapour Explosions (BLEVEs) due to accidents of vehicles transporting hazardous goods inside the tunnels. However, very limited study has investigated the influence of BLEVEs inside a tunnel on its dynamic response. The present study numerically investigates the dynamic response of an arched tunnel subjected to an internal BLEVE by using the software LS-DYNA. The BLEVE load is directly simulated by using commercial code FLACS or approximated by the commonly used TNT equivalency method to investigate the influences of loading predictions on the tunnel responses. The results show that the corner and upper arc of arched lining are more prone to be damaged under the directly predicted BLEVE loads. Compared to the directly predicted BLEVE loads, explosion loads estimated by the TNT equivalency method induce more significant damage to the lining due to its higher peak pressure, but smaller peak displacements since the corresponding impulse is lower than that of the BLEVE load. In addition, parametric studies are conducted to investigate the effects of concrete grade, concrete thickness, steel reinforcement ratio, and stiffness of surrounding rock mass on dynamic response of the arched tunnel subjected to the internal BLEVE. It is found that increasing concrete grade and thickness and enhancing the stiffness of surrounding rock mass are more effective than increasing steel reinforcement ratio in improving the performance of the arched tunnel against internal BLEVE loadings.

Due to global climate change–induced shifts in species distributions, estimating changes in community composition through the use of Species Distribution Models has become a key management tool. Being able to determine how species associations change along environmental gradients is likely to be pivotal in exploring the magnitude of future changes in species’ distributions. This is particularly important in connectivity-limited ecosystems, such as freshwater ecosystems, where increased human translocation is creating species associations over previously unseen environmental gradients. Here, we use a large-scale presence–absence dataset of freshwater fish from lakes across the Fennoscandian region in a Joint Species Distribution Model, to measure the effect of temperature on species associations. We identified a trend of negative associations between species tolerant of cold waters and those tolerant of warmer waters, as well as positive associations between several more warm-tolerant species, with these associations often shifting depending on local temperatures. Our results confirm that freshwater ecosystems can expect to see a large-scale shift towards communities dominated by more warm-tolerant species. While there remains much work to be done to predict exactly where and when local extinctions may take place, the model implemented provides a starting-point for the exploration of climate-driven community trends. This approach is especially informative in regards to determining which species associations are most central in shaping future community composition, and which areas are most vulnerable to local extinctions.

Population ageing is considered to be one of the most significant social phenomena that is transforming economies and societies around the world. According to the World Health Organization (WHO), ageing is on the rise. In Colombia, demographic growth exhibits a natural increase, which shows a notable difference between birth and general mortality rates. According to DANE, in Colombia, natural growth rates denote a precipitous decline over time. The Central and local governments can help with decision-making in order to establish sexual and reproductive health policies. Machine Learning (ML) therefore appears as a support tool, in which there are algorithms that allow creating models to learn from data and identify patterns that aid in supporting government entities in the decision-making process. Based on the above, this work proposes a method for ensemble ML algorithms, which supports decision-making regarding demographic control focused on birth. The prediction method made it possible to show that the decrease in births in Colombia in recent years is due to the change in the priorities of women and men. Women face discrimination and difficulty in accessing and staying in employment due to maternity. Consequently, it is difficult for them to articulate their professional life with the job market. Women have to assume a disproportionate burden of care, which is why they want to have fewer children, namely one or two at most., El envejecimiento de la población es considerado uno de los fenómenos sociales más significativos que está transformando las economías y las sociedades en todo el mundo. Según la Organización Mundial de la Salud (OMS) el envejecimiento está en aumento. En Colombia, el crecimiento demográfico presenta un incremento natural que muestra una notable diferencia entre las tasas de natalidad y las de mortalidad general. Según el DANE, en Colombia, las tasas de crecimiento natural denotan un vertiginoso declive a lo largo del tiempo. El gobierno central y los gobiernos locales pueden ayudar en la toma de decisiones para establecer políticas de salud sexual y reproductiva. Como herramienta de apoyo aparece el Machine Learning (ML), en el cual existen algoritmos que permiten crear modelos para aprender de los datos e identificar patrones que sirven para apoyar a entes gubernamentales en el proceso de toma de decisiones. Con base en lo anterior, este trabajo propone un método de ensamble de algoritmos de ML que apoye la toma de decisiones respecto al control demográfico enfocado en natalidad. El método de predicción permitió evidenciar que la disminución de nacimientos en Colombia durante los últimos años se debe al cambio en las prioridades de mujeres y hombres. Las mujeres enfrentan discriminación y dificultad en el acceso y la permanencia del empleo a causa de la maternidad. Como consecuencia, se les dificulta articular su vida profesional con el mercado laboral. Las mujeres tienen que asumir una carga desproporcionada de cuidado, por la cual quieren tener menos hijos, es decir uno o máximo dos.

Population ageing is considered to be one of the most significant social phenomena that is transforming economies and societies around the world. According to the World Health Organization (WHO), ageing is on the rise. In Colombia, demographic growth exhibits a natural increase, which shows a notable difference between birth and general mortality rates. According to DANE, in Colombia, natural growth rates denote a precipitous decline over time. The Central and local governments can help with decision-making in order to establish sexual and reproductive health policies. Machine Learning (ML) therefore appears as a support tool, in which there are algorithms that allow creating models to learn from data and identify patterns that aid in supporting government entities in the decision-making process. Based on the above, this work proposes a method for ensemble ML algorithms, which supports decision-making regarding demographic control focused on birth. The prediction method made it possible to show that the decrease in births in Colombia in recent years is due to the change in the priorities of women and men. Women face discrimination and difficulty in accessing and staying in employment due to maternity. Consequently, it is difficult for them to articulate their professional life with the job market. Women have to assume a disproportionate burden of care, which is why they want to have fewer children, namely one or two at most., El envejecimiento de la población es considerado uno de los fenómenos sociales más significativos que está transformando las economías y las sociedades en todo el mundo. Según la Organización Mundial de la Salud (OMS) el envejecimiento está en aumento. En Colombia, el crecimiento demográfico presenta un incremento natural que muestra una notable diferencia entre las tasas de natalidad y las de mortalidad general. Según el DANE, en Colombia, las tasas de crecimiento natural denotan un vertiginoso declive a lo largo del tiempo. El gobierno central y los gobiernos locales pueden ayudar en la toma de decisiones para establecer políticas de salud sexual y reproductiva. Como herramienta de apoyo aparece el Machine Learning (ML), en el cual existen algoritmos que permiten crear modelos para aprender de los datos e identificar patrones que sirven para apoyar a entes gubernamentales en el proceso de toma de decisiones. Con base en lo anterior, este trabajo propone un método de ensamble de algoritmos de ML que apoye la toma de decisiones respecto al control demográfico enfocado en natalidad. El método de predicción permitió evidenciar que la disminución de nacimientos en Colombia durante los últimos años se debe al cambio en las prioridades de mujeres y hombres. Las mujeres enfrentan discriminación y dificultad en el acceso y la permanencia del empleo a causa de la maternidad. Como consecuencia, se les dificulta articular su vida profesional con el mercado laboral. Las mujeres tienen que asumir una carga desproporcionada de cuidado, por la cual quieren tener menos hijos, es decir uno o máximo dos.

We present observations of [N II] 205 mu m, [O III]88 mu m, and dust emission in a stronglylensed, submillimeter galaxy (SMG) at z= 6.0, G09.83808, with the Atacama Large Millimeter/submillimeter Array (ALMA). Both [N II] and [O III] line emissions are detected at >12 sigma in the 0".8-resolution maps. Lens modeling indicates that the spatial distribution of the dust continuum emission is well characterized by a compact disk with an effective radius of 0.64 +/- 0.02 kpc and a high infrared surface brightness of Sigma(IR) = (1.8 +/- 0.3) x 10(12) L-circle dot kpc(-2). This result supports that G09.83808 is the progenitor of compact quiescent galaxies at z similar to 4, where the majority of its stars are expected to be formed through a strong and short burst of star formation. G09.83808 and other lensed SMGs show a decreasing trend in the [N Id line to infrared luminosity ratio with increasing continuum flux density ratio between 63 and 158 mu m, as seen in local luminous infrared galaxies (LIRGs). The decreasing trend can be reproduced by photoionization models with increasing ionization parameters. Furthermore, by combining the [N II]/[O III] luminosity ratio with far-infrared continuum flux density ratio in G09.83808, we infer that the gas phase metallicity is already Z approximate to 0.5-0.7 Z(circle dot). G09.83808 is likely one of the earliest galaxies that has been chemically enriched at the end of reionization.

Airlines are critical today for carrying people and commodities on time. Any delay in the schedule of these planes can potentially disrupt the business and trade of thousands of employees at any given time. Therefore, precise flight delay prediction is beneficial for the aviation industry and passenger travel. Recent research has focused on using artificial intelligence algorithms to predict the possibility of flight delays. Earlier prediction algorithms were designed for a specific air route or airfield. Many present flight delay prediction algorithms rely on tiny samples and are challenging to understand, allowing almost no room for machine learning implementation. This research study develops a flight delay prediction system by analyzing data from domestic flights inside the United States of America. The proposed models learn about the factors that cause flight delays and cancellations and the link between departure and arrival delays.

Species Distribution Models (SDMs) are essential tools for predicting climate change impact on species’ distributions and are commonly employed as an informative tool on which to base management and conservation actions. Focusing only on a part of the entire distribution of a species for fitting SDMs is a common approach. Yet, geographically restricting their range can result in considering only a subset of the species’ ecological niche (i.e.,niche truncation) which could lead to biased spatial predictions of future climate change effects, particularly if future conditions belong to those parts of the species ecological niche that have been excluded for model fitting. The integration of large-scale distribution data encompassing the whole species range with more regional data can improve future predictions but comes along with challenges owing to the broader scale and/or lower quality usually associated with these data. Here, we compare future predictions obtained from a traditional SDM fitted on a regional dataset (Switzerland) to predictions obtained from data integration methods that combine regional and European datasets for several bird species breeding in Switzerland. Three models were fitted: a traditional SDM based only on regional data and thus not accounting for niche truncation, a data pooling model where the two datasets are merged without considering differences in extent or resolution, and a downscaling hierarchical approach that accounts for differences in extent and resolution. Results show that the traditional model leads to much larger predicted range changes (either positively or negatively) under climate change than both data integration methods. The traditional model also identified different variables as main drivers of species’ distribution compared to data-integration models. Differences between models regarding predicted range changes were larger for species where future conditions were outside the range of conditions existing in the regiona

Airlines are critical today for carrying people and commodities on time. Any delay in the schedule of these planes can potentially disrupt the business and trade of thousands of employees at any given time. Therefore, precise flight delay prediction is beneficial for the aviation industry and passenger travel. Recent research has focused on using artificial intelligence algorithms to predict the possibility of flight delays. Earlier prediction algorithms were designed for a specific air route or airfield. Many present flight delay prediction algorithms rely on tiny samples and are challenging to understand, allowing almost no room for machine learning implementation. This research study develops a flight delay prediction system by analyzing data from domestic flights inside the United States of America. The proposed models learn about the factors that cause flight delays and cancellations and the link between departure and arrival delays.

Pendant longtemps, seul un rôle d’intermédiaire était connu à l’ARN. Cependant, il a été trouvé que grâce à sa forme simple brin, cette molécule peut se replier sur elle-même et ainsi constituer des structures qui lui confèrent diverses fonctions. Parmi ces formations, le G- quadruplexe (G4) est composé d’appariements de type Hoogsteen au lieu des appariements de type Watson-Crick. Les G4 peuvent se replier dans l’ARN, mais aussi dans l’ADN et y avoir de multiples rôles. Leur étude, initialement timide, a vécu un engouement dans les dernières années après avoir été découvertes dans les télomères et ayant un impact sur les cancers. Les G4 sont aussi fortement enrichis dans des régions régulatrices, comme les promoteurs dans les génomes ou les régions non traduites dans les transcrits. Quelques recherches laissent penser qu’il en est de même pour tous les eucaryotes, mais pas pour les bactéries. À ce jour, l’évolution des G4 reste peu étudiée, principalement à cause de la mauvaise conservation des séquences. Dans la présente thèse, la distribution des G4 dans les différents royaumes du vivant a été explorée, mais aussi différentes approches pour mieux appréhender leur évolution. Pour ce faire, une première étape était de prédire les G4 dans le transcriptome humain, et de valider ces prédictions. Ensuite, cette stratégie a été appliquée à diverses espèces au travers des trois royaumes du vivant. Ces données ont été analysées puis mises à disposition pour la communauté de chercheurs dans une base de données nommées GAIA. Finalement, une approche a été développée pour tenter de mieux comprendre l’évolution des G4., For a long time, RNA only known function was its intermediate role. Yet, it has been discovered that, thanks to its single-strand form, it can fold onto itself to generate structures and thus acquire diverse functions. Among these structures stands G-quadruplexes (G4s), which are made of Hoogsteen base-pairing instead of Watson & Crick ones. This structure can fold in RNA, but also in DNA where they can have multiple functions. The G4 study remained shy until the past few years when G4s interest suddenly increased with the discovery of their presence in telomeres and the impact they have on cancer. G4s are also highly enriched in other regulatory regions like promoters in genomes or in untranslated regions of transcripts. Some studies are implying that this distribution might be similar among all eukaryotes but not bacteria. The evolution of G4s is still relatively unknown and few studies are made on it. This is mainly due to one phenomenon: G4s sequences are badly conserved, which harden the task of its study. In the present thesis, it is proposed to study more in-depth G4s distribution among living kingdoms, but also to try new approaches to apprehend their evolution. To do so, the first step was to predict G4s inside the human transcriptome and to validate these predictions. Then, this method has been applied to diverse species among the three living kingdoms. These data have been be analysed and made public thanks to the GAIA database. Finally, an approach has been be developed to look at the evolution of G4s.

Recent developments in \ac{ML} have paved the way for unprecedented possibilities in the field of data analytics in numerous team sports, such as American football, baseball, and basketball. In more recent years, \ac{ML} techniques have been applied to football as professional teams got inspired to collect enormous quantities of data to evaluate the performance of their players and their applied tactics. However, research in performance analytics has primarily focused on the development of offline evaluation techniques, while online evaluation of a game has remained largely unexplored. To attract the attention of executive decision-makers in football, research requires the creation of comprehensive metrics that can give answers to the continuous calls of coaches and trainers to evaluate within-game performance. This work therefore aimed at finding an online evaluation technique to capture the complex aspects and highly non-linear dynamics of the sport in an interpretable way. \\ To achieve this goal, an online predictive model was developed using an \ac{ANN} approach, which is able to give an instantaneous value to every moment in the game through the computation of the \ac{ED}. The \ac{ED} is a comprehensive metric created in this work, which defines a form of relative, goal-related danger and encapsulates the complex spatio-temporal characteristics of the 22 players on the pitch. The \acp{ANN} have been trained and evaluated on two newly created datasets of simulated data which contain a combined total of 72 games and include the tracking data of the players and ball as well information on in-game events such as shots, passes, and goals. The extensiveness of the datasets enabled the creation of a novel \ac{MDP} that describes the complicated dynamics of the system. The developed \ac{MDP} model is a crucial component of this work and served as the basis for the design of the \ac{ED} and allowed the creation of the \ac{ANN}-based predict, Mechanical Engineering | Systems and Control

Structure predictions have matched the accuracy of experimental structures from close homologues, providing suitable models for molecular replacement phasing. Even in predictions that present large differences due to the relative movement of domains or poorly predicted areas, very accurate regions tend to be present. These are suitable for successful fragment-based phasing as implemented in ARCIMBOLDO. The particularities of predicted models are inherently addressed in the new predicted_model mode, rendering preliminary treatment superfluous but also harmless. B-value conversion from predicted LDDT or error estimates, the removal of unstructured polypeptide, hierarchical decomposition of structural units from domains to local folds and systematically probing the model against the experimental data will ensure the optimal use of the model in phasing. Concomitantly, the exhaustive use of models and stereochemistry in phasing, refinement and validation raises the concern of crystallographic model bias and the need to critically establish the information contributed by the experiment. Therefore, in its predicted_model mode ARCIMBOLDO_SHREDDER will first determine whether the input model already constitutes a solution or provides a straightforward solution with Phaser. If not, extracted fragments will be located. If the landscape of solutions reveals numerous, clearly discriminated and consistent probes or if the input model already constitutes a solution, model-free verification will be activated. Expansions with SHELXE will omit the partial solution seeding phases and all traces outside their respective masks will be combined in ALIXE, as far as consistent. This procedure completely eliminates the molecular replacement search model in favour of the inferences derived from this model. In the case of fragments, an incorrect starting hypothesis impedes expansion. The predicted_model mode has been tested in different scenarios.

We present observations of [N II] 205 mu m, [O III]88 mu m, and dust emission in a stronglylensed, submillimeter galaxy (SMG) at z= 6.0, G09.83808, with the Atacama Large Millimeter/submillimeter Array (ALMA). Both [N II] and [O III] line emissions are detected at >12 sigma in the 0".8-resolution maps. Lens modeling indicates that the spatial distribution of the dust continuum emission is well characterized by a compact disk with an effective radius of 0.64 +/- 0.02 kpc and a high infrared surface brightness of Sigma(IR) = (1.8 +/- 0.3) x 10(12) L-circle dot kpc(-2). This result supports that G09.83808 is the progenitor of compact quiescent galaxies at z similar to 4, where the majority of its stars are expected to be formed through a strong and short burst of star formation. G09.83808 and other lensed SMGs show a decreasing trend in the [N Id line to infrared luminosity ratio with increasing continuum flux density ratio between 63 and 158 mu m, as seen in local luminous infrared galaxies (LIRGs). The decreasing trend can be reproduced by photoionization models with increasing ionization parameters. Furthermore, by combining the [N II]/[O III] luminosity ratio with far-infrared continuum flux density ratio in G09.83808, we infer that the gas phase metallicity is already Z approximate to 0.5-0.7 Z(circle dot). G09.83808 is likely one of the earliest galaxies that has been chemically enriched at the end of reionization.

5G is currently being implemented around the world. A way to save resources in 5G could be to have several sector carriers sharing one power source. This requires being able to predict the electrical power of the sector carriers to make sure they do not exceed the capability of the power source. This thesis investigates deep learning for predicting the electrical power of a sector carrier. The chosen deep learning models are Multilayer perceptron (MLP) and Long short-term memory network (LSTM) and they are trained and evaluated on seven generated datasets from a sector carrier in Lund. The study consists of two parts, comparing the predictive performance between MLP and LSTM as well as determining the importance of each input feature on the predictions. It is concluded from the results that the MLP outperforms the LSTM in all datasets and should be the preferred model. The most important input feature for the predictions is by far the number of Resource elements (REs) used per slot. The number of Physical resource blocks (PRBs), modulation, scaling, number of streams and codebook indices have varying importance. The results can be used for guidance when implementing a real time system for which model and which input features to include. More work is required in investigating how the models perform on aggregated data from several sector carriers and determining if the models could feasibly be implemented for real time predictions., 5G håller på att implementeras runtom i världen. Ett sätt att spara resurser i 5G skulle kunna vara att ha flera sektorbärare dela på en strömkälla. För att implementera detta behöver man kunna förutsäga effekten av sektorbärarna för att försäkra sig om att de inte kommer överstiga den maximala effekten strömkällan tillåter. I denna avhandling undersöks djupinlärning för att förutsäga effekten av en sektorbärare. De använda djupinlärningsmodellerna är Multilayer perceptron (MLP) och Long short-term memory (LSTM) och de tränas och utvärderas på sju stycken olika dataset genererade från en sektorbärare i Lund. Studien är uppdelad i två delar, en del där MLP och LSTM jämförs med hänsyn till deras prediktiva förmåga och en del där varje datavariabel utvärderas till hur viktig den är för modellernas förutsägelser. Från resultaten kunde slutsatsen dras att MLP presterar bättre än LSTM på alla sju dataset och att MLP bör väljas före LSTM. Den viktigaste datavariabeln är antalet resurs element med en signifikant skillnad jämfört med övriga datavariabler. Antalet fysiska resursblock, modulationen, skalningsfaktorn, antalet strömmar och kodboksindex har varierande viktighet. Resultaten kan användas som stöd för implementering av realtid system för att bestämma lämplig model och data. Mer arbete krävs för att undersöka hur modellerna presterar på data som är aggregerat från flera sektorbärare samt undersöka huruvida det är genomförbart att implementera ett realtid system för förutsägelser utifrån beräkningskraft och kostandsanalys.

5G is currently being implemented around the world. A way to save resources in 5G could be to have several sector carriers sharing one power source. This requires being able to predict the electrical power of the sector carriers to make sure they do not exceed the capability of the power source. This thesis investigates deep learning for predicting the electrical power of a sector carrier. The chosen deep learning models are Multilayer perceptron (MLP) and Long short-term memory network (LSTM) and they are trained and evaluated on seven generated datasets from a sector carrier in Lund. The study consists of two parts, comparing the predictive performance between MLP and LSTM as well as determining the importance of each input feature on the predictions. It is concluded from the results that the MLP outperforms the LSTM in all datasets and should be the preferred model. The most important input feature for the predictions is by far the number of Resource elements (REs) used per slot. The number of Physical resource blocks (PRBs), modulation, scaling, number of streams and codebook indices have varying importance. The results can be used for guidance when implementing a real time system for which model and which input features to include. More work is required in investigating how the models perform on aggregated data from several sector carriers and determining if the models could feasibly be implemented for real time predictions., 5G håller på att implementeras runtom i världen. Ett sätt att spara resurser i 5G skulle kunna vara att ha flera sektorbärare dela på en strömkälla. För att implementera detta behöver man kunna förutsäga effekten av sektorbärarna för att försäkra sig om att de inte kommer överstiga den maximala effekten strömkällan tillåter. I denna avhandling undersöks djupinlärning för att förutsäga effekten av en sektorbärare. De använda djupinlärningsmodellerna är Multilayer perceptron (MLP) och Long short-term memory (LSTM) och de tränas och utvärderas på sju stycken olika dataset genererade från en sektorbärare i Lund. Studien är uppdelad i två delar, en del där MLP och LSTM jämförs med hänsyn till deras prediktiva förmåga och en del där varje datavariabel utvärderas till hur viktig den är för modellernas förutsägelser. Från resultaten kunde slutsatsen dras att MLP presterar bättre än LSTM på alla sju dataset och att MLP bör väljas före LSTM. Den viktigaste datavariabeln är antalet resurs element med en signifikant skillnad jämfört med övriga datavariabler. Antalet fysiska resursblock, modulationen, skalningsfaktorn, antalet strömmar och kodboksindex har varierande viktighet. Resultaten kan användas som stöd för implementering av realtid system för att bestämma lämplig model och data. Mer arbete krävs för att undersöka hur modellerna presterar på data som är aggregerat från flera sektorbärare samt undersöka huruvida det är genomförbart att implementera ett realtid system för förutsägelser utifrån beräkningskraft och kostandsanalys.

The global decline of water quality in rivers and streams has resulted in a pressing need to design new watershed management strategies. Water quality can be affected by multiple stressors including population growth, land use change, global warming, and extreme events, with repercussions on human and ecosystem health. A scientific understanding of factors affecting riverine water quality and predictions at local to regional scales, and at sub-daily to decadal timescales are needed for optimal management of watersheds and river basins. Here, we discuss how machine learning (ML) can enable development of more accurate, computationally tractable, and scalable models for analysis and predictions of river water quality. We review relevant state-of-the art applications of ML for water quality models and discuss opportunities to improve the use of ML with emerging computational and mathematical methods for model selection, hyperparameter optimization, incorporating process knowledge into ML models, improving explainablity, uncertainty quantification, and model-data integration. We then present considerations for using ML to address water quality problems given their scale and complexity, available data and computational resources, and stakeholder needs. When combined with decades of process understanding, interdisciplinary advances in knowledge-guided ML, information theory, data integration, and analytics can help address fundamental science questions and enable decision-relevant predictions of riverine water quality.

Airlines are critical today for carrying people and commodities on time. Any delay in the schedule of these planes can potentially disrupt the business and trade of thousands of employees at any given time. Therefore, precise flight delay prediction is beneficial for the aviation industry and passenger travel. Recent research has focused on using artificial intelligence algorithms to predict the possibility of flight delays. Earlier prediction algorithms were designed for a specific air route or airfield. Many present flight delay prediction algorithms rely on tiny samples and are challenging to understand, allowing almost no room for machine learning implementation. This research study develops a flight delay prediction system by analyzing data from domestic flights inside the United States of America. The proposed models learn about the factors that cause flight delays and cancellations and the link between departure and arrival delays.

New Student Admission is one of the essential activities carried out regularly every year or semester. As the operational system of student admissions progresses, student admission data increases yearly. ESA Unggul University (UEU) has not used this data to make strategic decisions, market potential, and consider invitations to enter the academic path. So it is necessary to conduct research whose results can be used by UEU in analyzing prospective students at the time of new student admissions. In this study, data analysis was carried out from 2014 to 2019. This study aims to produce a design using the classification method to predict whether prospective students are accepted or withdrawn. In this study, 19,603 training data and 4,901 test data were used. The results showed the best Random Forest algorithm with an accuracy of 73.61%. The results of this study can be used to support the marketing department in minimizing the number of prospective students who resign.

Species Distribution Models (SDMs) are essential tools for predicting climate change impact on species’ distributions and are commonly employed as an informative tool on which to base management and conservation actions. Focusing only on a part of the entire distribution of a species for fitting SDMs is a common approach. Yet, geographically restricting their range can result in considering only a subset of the species’ ecological niche (i.e., niche truncation) which could lead to biased spatial predictions of future climate change effects, particularly if future conditions belong to those parts of the species ecological niche that have been excluded for model fitting. The integration of large-scale distribution data encompassing the whole species range with more regional data can improve future predictions but comes along with challenges owing to the broader scale and/or lower quality usually associated with these data. Here, we compare future predictions obtained from a traditional SDM fitted on a regional dataset (Switzerland) to predictions obtained from data integration methods that combine regional and European datasets for several bird species breeding in Switzerland. Three models were fitted: a traditional SDM based only on regional data and thus not accounting for niche truncation, a data pooling model where the two datasets are merged without considering differences in extent or resolution, and a downscaling hierarchical approach that accounts for differences in extent and resolution. Results show that the traditional model leads to much larger predicted range changes (either positively or negatively) under climate change than both data integration methods. The traditional model also identified different variables as main drivers of species’ distribution compared to data-integration models. Differences between models regarding predicted range changes were larger for species where future conditions were outside the range of conditions existing in the region

In the last decade, artificial intelligence (AI) has revolutionised the field of drug research. Staff abilities (55 percent), data structure (52 percent), and resources were all factors in AI deployment (49 percent ). Nearly 60% of respondents said they expected to hire more people in the next two years to assist AI usage or adoption in drug development. AI in areas like as drug research and development, drug repurposing, boosting pharmaceutical productivity, and clinical trials, among others, minimises human effort and allows for the achievement of objectives in a short amount of time. On the one hand, AI techniques used in drug development bring the drug development process and the use of various models closer to medicinal chemists, while on the other hand, AI methods used in drug development bring the drug development process and the use of various models closer to mathematicians.

In the last decade, artificial intelligence (AI) has revolutionised the field of drug research. Staff abilities (55 percent), data structure (52 percent), and resources were all factors in AI deployment (49 percent ). Nearly 60% of respondents said they expected to hire more people in the next two years to assist AI usage or adoption in drug development. AI in areas like as drug research and development, drug repurposing, boosting pharmaceutical productivity, and clinical trials, among others, minimises human effort and allows for the achievement of objectives in a short amount of time. On the one hand, AI techniques used in drug development bring the drug development process and the use of various models closer to medicinal chemists, while on the other hand, AI methods used in drug development bring the drug development process and the use of various models closer to mathematicians.

The global decline of water quality in rivers and streams has resulted in a pressing need to design new watershed management strategies. Water quality can be affected by multiple stressors including population growth, land use change, global warming, and extreme events, with repercussions on human and ecosystem health. A scientific understanding of factors affecting riverine water quality and predictions at local to regional scales, and at sub-daily to decadal timescales are needed for optimal management of watersheds and river basins. Here, we discuss how machine learning (ML) can enable development of more accurate, computationally tractable, and scalable models for analysis and predictions of river water quality. We review relevant state-of-the art applications of ML for water quality models and discuss opportunities to improve the use of ML with emerging computational and mathematical methods for model selection, hyperparameter optimization, incorporating process knowledge into ML models, improving explainablity, uncertainty quantification, and model-data integration. We then present considerations for using ML to address water quality problems given their scale and complexity, available data and computational resources, and stakeholder needs. When combined with decades of process understanding, interdisciplinary advances in knowledge-guided ML, information theory, data integration, and analytics can help address fundamental science questions and enable decision-relevant predictions of riverine water quality.

Recent developments in \ac{ML} have paved the way for unprecedented possibilities in the field of data analytics in numerous team sports, such as American football, baseball, and basketball. In more recent years, \ac{ML} techniques have been applied to football as professional teams got inspired to collect enormous quantities of data to evaluate the performance of their players and their applied tactics. However, research in performance analytics has primarily focused on the development of offline evaluation techniques, while online evaluation of a game has remained largely unexplored. To attract the attention of executive decision-makers in football, research requires the creation of comprehensive metrics that can give answers to the continuous calls of coaches and trainers to evaluate within-game performance. This work therefore aimed at finding an online evaluation technique to capture the complex aspects and highly non-linear dynamics of the sport in an interpretable way. \\ To achieve this goal, an online predictive model was developed using an \ac{ANN} approach, which is able to give an instantaneous value to every moment in the game through the computation of the \ac{ED}. The \ac{ED} is a comprehensive metric created in this work, which defines a form of relative, goal-related danger and encapsulates the complex spatio-temporal characteristics of the 22 players on the pitch. The \acp{ANN} have been trained and evaluated on two newly created datasets of simulated data which contain a combined total of 72 games and include the tracking data of the players and ball as well information on in-game events such as shots, passes, and goals. The extensiveness of the datasets enabled the creation of a novel \ac{MDP} that describes the complicated dynamics of the system. The developed \ac{MDP} model is a crucial component of this work and served as the basis for the design of the \ac{ED} and allowed the creation of the \ac{ANN}-based predict, Mechanical Engineering | Systems and Control

In dieser Arbeit werden Stornierungen von Hotelbuchungen mithilfe verschiedener Methoden maschinellen Lernens vorhergesagt. Buchungsstornierungen stellen für die Hotelindustrie ein Problem dar, da sie die Planbarkeit von Einnahmen und Kosten stark erschweren. Unter Verwendung von logistischen Regressionen, Random Forests und künstlichen neuronalen Netzen werden Modelle erstellt, die diese Stornierungen vorhersagen sollen. Diese verschiedenen Modelle werden anschließend miteinander verglichen, um die effektivste Methode zu finden. Vorhersagemodelle könnten bestehende Probleme, die durch Stornierungen entstehen, lösen, in dem Maßnahmen wie Überbuchungen effektiver eingesetzt werden könnten. Bei der Analyse wird ein Datensatz, der von zwei portugiesischen Hotels stammt, mit fast 120.000 realen Beobachtungen und 31 Variablen zu Hilfe genommen., The aim of this thesis is to predict hotel booking cancellations with different machine learning methods. Hotel booking cancellations are a problem for hotels, because they diminish revenue and make it harder to plan future income. Using logistic regressions, Random Forests and artificial neural networks different prediction models are computed. The models are being compared to eachother to find the most effective method. These models could solve existing problems caused by cancellation, because measures such as overbooking could be used more efficiently. The analysis is based on a real dataset from two portuguese hotels and has almost 120,000 observations and 31 variables.

Predictive modelling of the fire duration, fire temperatures, heat release rates and the structural capacity during building fires can be used to show compliance with performance-based building code requirements. The predictive models presented in this report focusses on the post flashover fire including the decay phase and extinction of flaming combustion for mass timber structures. A priori predictions of five recent compartment fire tests have been set against experimental results and compared. After the tests, the model has been updated, mostly for increased ease of use and increased accuracy for the decay phase. The model consists of a single-zone model which uses an energy equilibrium approach to obtain gas temperatures and surface temperatures of compartment boundaries. The energy contribution of charring mass timber is included using through-depth temperature calculations of the structure and experimental relationships to determine the combustion rate. The through-depth temperatures of mass timber members also serve to provide information for structural calculations using temperature dependent reduced material properties. However, the structural calculations are out of the scope of the current report. The radiation conditions (and total thermal exposure to walls ceilings and floors) predicted by the updated model were accurately described the of recent full-scale experiments within the variations between and within the tests. The comparisons with experiments showed that the total heat is, however, underestimated in some cases and surface temperatures were underestimated in the decay phase. Local effects caused by a radiative feedback loop between surfaces that show significant char oxidation, which occurred in a part of the test, is not included in the model.

Forecasting using judgment is common in practice. From project planning to investing in a house, we routinely make important decisions based on how we expect the future to unfold. In much the same way, governments and businesses rely on forecasts related to political and economic events in order to appropriately plan for future challenges and opportunities. The cybersecurity industry is no different, and each year many security organizations release ``annual predictions'' of anticipated cybersecurity issues. The premise behind such predictions is that experts are in a superior position due to their knowledge and experience to anticipate future cybersecurity-related events. However, this premise is untested and motivates this research.This master's thesis examines the primary question of whether security professionals' predictions about future cybersecurity-related questions are systematically distinct from those of other Information Technology professionals, who lack the same specialized experience. In particular, this research examines the results of 20 security-related surveys. Each survey takes a different approach to the category and format of the question in order to analyze a variety of types of forecasting questions. Using a combination of measurement techniques, I determine if there exist any significant patterns among security professionals and non-security professionals. Additionally, I analyze two cohorts within security professionals to determine if there are measurable differences between them with regard to judgment forecasts. Through this study, I concluded that I cannot support the claim that security professionals offer a distinct judgment over non-security experts.

The subject of forecasting earthquakes is an intriguing one to investigate. As a natural calamity, earthquakes continue to be devastating, not just to the economy but also to the lives of individuals. This gave rise to the concept of creating an early warning system against seismic catastrophes to minimize deaths. Researchers have been making earthquake forecasts and seismic hazard ratings of a location for a few years now. In this work, we attempt to forecast earthquakes before they occur using p-arrival data, which includes information on disaster arrival time and amplitude height from the arrival station. Several studies on earthquake prediction have been carried out so far and have developed and used the Random Forest method and one of the Machine Learning. According to [1], the process of predicting earthquakes has been studied for a long time, but there is still uncertainty due to the diversity and complexity of the earthquake phenomenon itself. According to [2], conducting a random forest prediction model to identify the structural safety status of buildings damaged by the earthquake is probabilistic. An earthquake's latitude, longitude, magnitude, and depth may be predicted using the random forest algorithm. A random forest with multioutput technique is employed, with variables being each station's recorded value and geographic position. This study's predictions were accurate to within 63 percent.

Aim Plant functional traits summarize the main variability in plant form and function across taxa and biomes. We assess whether geographic range size, climatic niche size, and local abundance of plants can be predicted by sets of traits (trait syndromes) or are driven by single traits. Location Eurasia. Methods Species distribution maps were extracted from the Chorological Database Halle to derive information on the geographic range size and climatic niche size for 456 herbaceous, dwarf shrub and shrub species. We estimated local species abundances based on 740,113 vegetation plots from the European Vegetation Archive, where abundances were available as plant species cover per plot. We compiled a complete species-by-trait matrix of 20 plant functional traits from trait databases (TRY, BiolFlor and CLO-PLA). The relationships of species' geographic range size, climatic niche size and local abundance with single traits and trait syndromes were tested with multiple linear regression models. Results Generally, traits were more strongly related to local abundances than to broad-scale species distribution patterns in geographic and climatic space (range and niche size), but both were better predicted by trait combinations than by single traits. Local abundance increased with leaf area and specific leaf area (SLA). Geographic range size and climatic niche size both increased with SLA. While range size increased with plant height, niche size decreased with leaf carbon content. Conclusion Functional traits matter for species' abundance and distribution at both local and broad geographic scale. Local abundances are associated with different combinations of traits as compared to broad-scale distributions, pointing to filtering by different environmental and ecological factors acting at distinct spatial scales. However, traits related to the leaf economics spectrum were important for species' abundance and occurrence at both spatial scales. This finding emphasizes the general i

IoT har blivit en stor producent av big data. Big data kan användas för att optimera operationer, för att kunna göra det så måste man kunna extrahera användbar information från big data. Detta kan göras med hjälp av neurala nätverk och maskininlärning, vilket kan leda till nya typer av smarta applikationer. Den här rapporten fokuserar på att besvara frågan hur bra är neurala nätverk på att förutspå sensor värden och hur pålitliga är förutsägelserna och om dom kan användas i verkliga applikationer. Sensorlådor användes för att samla data från olika rum och olika neurala nätverksmodeller baserade på LSTM nätverk användes för att förutspå framtida värden. Dessa värden jämfördes sedan med dom riktiga värdena och absoluta medelfelet och standardavvikelsen beräknades. Tiden som behövdes för att producera en förutsägelse mättes och medelvärde och standardavvikelsen beräknades även där. LSTM modellerna utvärderades utifrån deras prestanda och träffsäkerhet. Modellen som endast förutspådde ett värde hade bäst träffsäkerhet, och modellerna tappade träffsäkerheten desto längre in i framtiden dom försökte förutspå. Resultaten visar att även dom enkla modellerna som skapades i detta projekt kan med säkerhet förutspå värden och därför användas i olika applikationer där extremt bra förutsägelser inte behövs., The IoT is becoming an increasing producer of big data. Big data can be used to optimize operations, realizing this depends on being able to extract useful information from big data. With the use of neural networks and machine learning this can be achieved and can enable smart applications that use this information. This thesis focuses on answering the question how good are neural networks at predicting sensor values and is the predictions reliable and useful in a real-life application? Sensory boxes were used to gather data from rooms, and several neural networks based on LSTM were used to predict the future values of the sensors. The absolute mean error of the predictions along with the standard deviation was calculated. The time needed to produce a prediction was measured as an absolute mean values with standard deviation. The LSTM models were then evaluated based on their performance and prediction accuracy. The single-step model, which only predicts the next timestep was the most accurate. The models loose accuracy when they need to predict longer periods of time. The results shows that simple models can predict the sensory values with some accuracy, while they may not be useful in areas where exact climate control is needed the models can be applicable in work areas such as schools or offices.

Aim Plant functional traits summarize the main variability in plant form and function across taxa and biomes. We assess whether geographic range size, climatic niche size, and local abundance of plants can be predicted by sets of traits (trait syndromes) or are driven by single traits. Location Eurasia. Methods Species distribution maps were extracted from the Chorological Database Halle to derive information on the geographic range size and climatic niche size for 456 herbaceous, dwarf shrub and shrub species. We estimated local species abundances based on 740,113 vegetation plots from the European Vegetation Archive, where abundances were available as plant species cover per plot. We compiled a complete species-by-trait matrix of 20 plant functional traits from trait databases (TRY, BiolFlor and CLO-PLA). The relationships of species' geographic range size, climatic niche size and local abundance with single traits and trait syndromes were tested with multiple linear regression models. Results Generally, traits were more strongly related to local abundances than to broad-scale species distribution patterns in geographic and climatic space (range and niche size), but both were better predicted by trait combinations than by single traits. Local abundance increased with leaf area and specific leaf area (SLA). Geographic range size and climatic niche size both increased with SLA. While range size increased with plant height, niche size decreased with leaf carbon content. Conclusion Functional traits matter for species' abundance and distribution at both local and broad geographic scale. Local abundances are associated with different combinations of traits as compared to broad-scale distributions, pointing to filtering by different environmental and ecological factors acting at distinct spatial scales. However, traits related to the leaf economics spectrum were important for species' abundance and occurrence at both spatial scales. This finding emphasizes the general i

One of the primary goals for the upcoming James Webb Space Telescope is to observe the first galaxies. Predictions for planned and proposed surveys have typically focused on average galaxy counts, assuming a random distribution of galaxies across the observed field. The first and most-massive galaxies, however, are expected to be tightly clustered, an effect known as cosmic variance. We show that cosmic variance is likely to be the dominant contribution to uncertainty for high-redshift mass and luminosity functions, and that median high-redshift and high-mass galaxy counts for planned observations lie significantly below average counts. Several different strategies are considered for improving our understanding of the first galaxies, including adding depth, area, and independent pointings. Adding independent pointings is shown to be the most efficient both for discovering the single highest-redshift galaxy and also for constraining mass and luminosity functions.

Fifty years ago the standard model offered an elegant new step towards understanding elementary fermion and boson fields, making several assumptions, suggested by experiments. The assumptions are still waiting for an explanation. There are many proposals in the literature for the next step. The spin-charge-family theory of one of us (N.S.M.B.) is offering the explanation for not only all by the standard model assumed properties of quarks and leptons and antiquarks and antileptons, with the families included, of the vectors gauge fields, of the Higgs scalar and Yukawa couplings, but also for the second quantization postulates of Dirac and for cosmological observations, like there are the appearance of the dark matter, of matter-antimatter asymmetry, making several predictions. This theory proposes a simple starting action in d >= (13 + 1)-dimensional space with fermions interacting with the gravity only (the vielbeins and the two kinds of the spin connection fields), what manifests in d = (3 + 1) as the vector and scalar gauge fields, and uses the odd Clifford algebra to describe the internal space of fermions, what enables that the creation and annihilation operators for fermions fulfil the anticommutation relations for the second quantized fields without Dirac's postulates: Fermions single particle states already anticommute. We present in this review article a short overview of the spin-charge-family theory, illustrating shortly on the toy model the breaks of the starting symmetries in d = (13+1)-dimensional space, which are triggered either by scalar fields - the vielbeins with the space index belonging to d > (3+1) - or by the condensate of the two right handed neutrinos, with the family quantum number not belonging to the observed families. We compare properties and predictions of this theory with the properties and predictions of SO(10) unifying theories. (C) 2021 Elsevier B.V. All rights reserved.

Predictive modelling of the fire duration, fire temperatures, heat release rates and the structural capacity during building fires can be used to show compliance with performance-based building code requirements. The predictive models presented in this report focusses on the post flashover fire including the decay phase and extinction of flaming combustion for mass timber structures. A priori predictions of five recent compartment fire tests have been set against experimental results and compared. After the tests, the model has been updated, mostly for increased ease of use and increased accuracy for the decay phase. The model consists of a single-zone model which uses an energy equilibrium approach to obtain gas temperatures and surface temperatures of compartment boundaries. The energy contribution of charring mass timber is included using through-depth temperature calculations of the structure and experimental relationships to determine the combustion rate. The through-depth temperatures of mass timber members also serve to provide information for structural calculations using temperature dependent reduced material properties. However, the structural calculations are out of the scope of the current report. The radiation conditions (and total thermal exposure to walls ceilings and floors) predicted by the updated model were accurately described the of recent full-scale experiments within the variations between and within the tests. The comparisons with experiments showed that the total heat is, however, underestimated in some cases and surface temperatures were underestimated in the decay phase. Local effects caused by a radiative feedback loop between surfaces that show significant char oxidation, which occurred in a part of the test, is not included in the model.

The latest progress in machine learning (ML) algorithms enabled to predict some steel physical properties previously modelled by linear regression (LR), such as the Ms temperature. Authors claimed that the performance given by ML models could improve the one of previous LR models, although they did not include fair comparisons. In this work, a large database was used to train different ML algorithms, whose Ms temperature predictions were compared to the ones of previous literature empirical models. ML methods were proved to require longer computational times and wider knowledge, while leading to similar results. Therefore, we recommend that ML methods are not always considered as the first option when trying to solve easy problems that can be modelled by LR techniques.

External conditions can drive biological rates in ectotherms by directly influencing body temperatures. While estimating the temperature dependence of performance traits such as growth and development rate is feasible under controlled laboratory settings, predictions in nature are difficult. One major challenge lies in translating performance under constant conditions to fluctuating environments. Using the butterfly Pieris napi as model system, we show that development rate, an important fitness trait, can be accurately predicted in the field using models parameterized under constant laboratory temperatures. Additionally, using a factorial design, we show that accurate predictions can be made across microhabitats but critically hinge on adequate consideration of non-linearity in reaction norms, spatial heterogeneity in microclimate and temporal variation in temperature. Our empirical results are also supported by a comparison of published and simulated data. Conclusively, our combined results suggest that, discounting direct effects of temperature, insect development rates are generally unaffected by thermal fluctuations.

Forecasting using judgment is common in practice. From project planning to investing in a house, we routinely make important decisions based on how we expect the future to unfold. In much the same way, governments and businesses rely on forecasts related to political and economic events in order to appropriately plan for future challenges and opportunities. The cybersecurity industry is no different, and each year many security organizations release ``annual predictions'' of anticipated cybersecurity issues. The premise behind such predictions is that experts are in a superior position due to their knowledge and experience to anticipate future cybersecurity-related events. However, this premise is untested and motivates this research.This master's thesis examines the primary question of whether security professionals' predictions about future cybersecurity-related questions are systematically distinct from those of other Information Technology professionals, who lack the same specialized experience. In particular, this research examines the results of 20 security-related surveys. Each survey takes a different approach to the category and format of the question in order to analyze a variety of types of forecasting questions. Using a combination of measurement techniques, I determine if there exist any significant patterns among security professionals and non-security professionals. Additionally, I analyze two cohorts within security professionals to determine if there are measurable differences between them with regard to judgment forecasts. Through this study, I concluded that I cannot support the claim that security professionals offer a distinct judgment over non-security experts.