[EN] Statistical Process Control (SPC) and the Automatic Process Control (APC) are two control philosophies have evolved independently until recently. The overall objective of both APC and SPC is to optimize the performance of processes, reducing the variability of the resulting characteristics around the desired values. The fundamentals of the two disciplines arise from the idea that the whole process has operation variations. These variations may affect, to a greater or lesser extent, the final product quality and the process productivity. The two methodologies conceptualize processes and control in different ways, they originated in different industrial sectors and have evolved independently, until the interest to integrate them in the control of industrial processes was deduced. It was warned that they could be complementary rather than conflicting methodologies as they were understood until then. The possibility of combining the advantages of both, integrating them into a new control paradigm was explored. First, the problem of identifying and estimating a process model is considered. Controlled variables in this model are the main feature of product quality and a productivity variable. The latter is innovative since the productivity variables are measured, but they are not considered as controlled variables. For this, two methods of multivariate time series are used, the Box-Jenkins multiple transfer function in the parsimonious way and the impulse response function obtained by Partial Least Squares regression (Time Series-Partial Least Squares, TS-PLS). These two methods were compared taking into account different aspects such as the simplicity of the modeling process in the stages of identification, estimation and model validation, as well as the utility of graphical tools that provide both methodologies, the goodness of fit obtained, and the simplicity of the mathematical structure of the model. The DMC (Dynamic Matrix Control) controller, an automatic control algorithm belonging to the family of MPC (Model Predictive Control) controllers, is derived from the estimated Box-Jenkins multiple transfer function that has been selected as the most suitable for this kind of processes. An optimal tuning method to maximize the controller performance, applying experimental design 2k-p, is presented. Finally, an integrated control system MESPC (Multivariate Engineering Statistical Process Control) whose monitoring component has been implemented applying latent structures based multivariate statistical process control methods (Lsb-MSPC), has been developed. The monitoring module is designed to act as both process and DMC controller supervisor. To do this, we estimate a NOC-PCA model (Normal Operation Conditions Principal Component Analysis), which has as variables both process-related and quality-related variables, all derived from the automatic control system. From this model, and DModX graphics have been derived. We assessed the performance of MESPC system, subjecting it to simulated potential failures or special causes of variability. / [ES] El Control Estadístico de Procesos (Statistical Process Control, SPC) y el Control Automático de Procesos (Automatic Process Control, APC) son dos filosofías de control se han desarrollado hasta recientemente de forma independiente. El objetivo general tanto del SPC como del APC, es optimizar el funcionamiento de los procesos, reduciendo la variabilidad de las características resultantes en torno a los valores deseados. El fundamento de ambas disciplinas, parte de la idea de que todo proceso presenta variaciones en su funcionamiento. Estas variaciones pueden afectar en mayor o en menor medida a la calidad final del producto y a la productividad del proceso. Las dos metodologías conceptualizan los procesos y su control de diferentes formas, se originaron en diferentes sectores industriales y han evolucionado de forma independiente, hasta que se dedujo el interés de integrarlas en el control de los procesos industriales, ya que se advirtió que podían ser complementarias, antes que contrapuestas, como se entendían hasta entonces y se exploró la posibilidad de aunar las ventajas de ambas, integrándolas en un nuevo paradigma de control. Esta tesis se centra en el estudio de la integración de procedimientos multivariantes para la regulación óptima y la monitorización estadística de procesos, con el propósito de contribuir a la mejora de la calidad y de la productividad de los procesos. La metodología propuesta se ha aplicado con fines ilustrativos a un proceso MIMO de producción en continuo de Polietileno de Alta Densidad (PEAD).En primer lugar, se considera el problema de la identificación y posterior estimación de un modelo del proceso. Las variables controladas en este modelo han sido la principal característica de calidad del producto y una variable de productividad, esto último es innovador puesto que las variables de productividad se miden, pero no se consideran variables controladas. Para ello, se emplean dos metodologías de series temporales multivariantes, la obtención de la función de transferencia múltiple en forma parsimoniosa de Box-Jenkins y la obtención de la función de respuesta a impulsos mediante los modelos de regresión por mínimos cuadrados parciales (Time Series-Partial Least Squares, TS-PLS). Estas dos metodologías se han comparado teniendo en cuenta distintos aspectos como son la simplicidad del proceso de modelado en las etapas de identificación, estimación y validación del modelo, así como la utilidad de las herramientas gráficas que proporcionan ambas metodologías, la bondad de ajuste obtenida, y la simplicidad de la estructura matemática del modelo. A partir del modelo de función de transferencia múltiple estimado, elegido como el más adecuado para este tipo de procesos, se desarrolla el controlador DMC (Dynamic Matrix Control), un algoritmo de control automático que pertenece a la familia del Control Predictivo basado en Modelos (Model Predictive Control, MPC). Se presenta un método de sintonizado óptimo del controlador que permita maximizar su rendimiento, aplicando diseño de experimentos 2k-p.Finalmente, se ha desarrollado un sistema de control integrado MESPC (Multivariate Engineering Statistical Process Control), cuya componente de monitorización se ha implementado aplicando métodos de control estadístico multivariante de procesos basados en técnicas de proyección en estructuras latentes. Este módulo de monitorización se ha diseñado para que actúe como supervisor tanto del proceso como del controlador DMC. Para ello, se ha estimado un modelo NOC-PCA (Normal Operation Conditions Principal Component Analysis), en el que han intervenido tanto variables relacionadas con el proceso como con la calidad, todas derivadas de la componente del control automático. A partir de este modelo se han derivado los gráficos y DModX. Se ha evaluado el funcionamiento del sistema MESPC, sometiéndolo a fallos potenciales o causas especiales de variabiliabilidad. / [CAT] El Control Estadístic de Processos (Statistical Process Control, SPC) i del Control Automàtic de Processos (Automatic Process Control, APC) son dues filosofies de control s'han desenvolupat fins a recentment de forma independent. L'objectiu general tant del SPC com del APC, és optimitzar el funcionament dels processos, reduint la variabilitat de les característiques resultants entorn dels valors desitjats. El fonament d'ambdues disciplines, part de la idea que tot procés presenta variacions en el seu funcionament. Aquestes variacions poden afectar en major o en menor mesura a la qualitat final del producte i a la productivitat del procés. Les dues metodologies conceptualitzen els processos i el seu control de diferents formes, es van originar en diferents sectors industrials i han evolucionat de forma independent, fins que es va deduir l'interès d'integrar-les en el control dels processos industrials, ja que es va advertir que podien ser complementàries, abans que contraposades, com s'entenien fins llavors i es va explorar la possibilitat de conjuminar els avantatges d'ambdues, integrant-les en un nou paradigma de control. Aquesta tesi se centra en l'estudi de la integració de procediments multivariants per a la regulació òptima i el monitoratge estadístic de processos amb el propòsit de contribuir a la millora de la qualitat i de la productivitat dels processos. La metodologia proposada s'ha aplicat amb finalitats il·lustratives a un procés MIMO de producció en continu de Polietilè d'Alta Densitat (PEAD). En primer lloc, es considera el problema de la identificació i posterior estimació d'un model del procés. Les variables controlades en aquest model han sigut la principal característica de qualitat del producte i una variable de productivitat, açò últim és innovador ja que les variables de productivitat es mesuren, però no es consideren variables controlades. Per a açò, s'utilitzen dues metodologies de sèries temporals multivariants, l'obtenció de la funció de transferència múltiple en forma parsimòniosa de Box-Jenkins i l'obtenció de la funció de resposta a impulsos mitjançant els models de regressió per mínims quadrats parcials (Times Series-Partial Least Squares, TS-PLS). Aquestes dues metodologies s'han comparat tenint en compte diferents aspectes com són la simplicitat del procés de modelatge en les etapes d'identificació, estimació i validació del model, així com la utilitat de les eines gràfiques que proporcionen ambdues metodologies, la bondat d'ajust obtinguda, i la simplicitat de l'estructura matemàtica del model. A partir del model de funció de transferència múltiple estimat, triat com el més adequat per a aquest tipus de processos, es desenvolupa el controlador DMC (Dynamic Matrix Control), un algorisme de control automàtic que pertany a la família del Control Predictiu basat en Models (Model Predictive Control, MPC). Es presenta un mètode de sintonitzat òptim del controlador que permeta maximitzar el seu rendiment, aplicant disseny d'experiments 2k-p. Finalment, s'ha desenvolupat un sistema de control integrat MESPC (Multivariate Engineering Statistical Process Control). Per a implementar la component de monitoratge d'aquest sistema integrat s'han usat mètodes de control estadístic multivariants de processos basats en tècniques de projecció en estructures latents (Latent structures based-Multivariate Statistical Process Control). Aquest mòdul de monitoratge s'ha dissenyat perquè actue com a supervisor tant del procés com del controlador DMC. Per a açò, s'ha estimat un model NOC-PCA (Normal Operation Conditions Principal Component Analysis), en el qual han intervingut variables relacionades tant amb el procés, com amb la qualitat, totes derivades de la component del control automàtic. A partir d'aquest model s'han derivat els gràfics i DModX. S'ha avaluat el funcionament del sistema MESPC, sotmetent-lo a fallades potencials o causes especials de / Barceló Cerdá, S. (2016). Estudio de la integración de procedimientos multivariantes para la regulación óptima y monitorización estadística de procesos [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/63442 / TESIS

Machine translation is a fundamental technology that is gaining more importance each day in our multilingual society. Companies and particulars are turning their attention to machine translation since it dramatically cuts down their expenses on translation and interpreting. However, the output of current machine translation systems is still far from the quality of translations generated by human experts. The overall goal of this thesis is to narrow down this quality gap by developing new methodologies and tools that improve the broader and more efficient deployment of machine translation technology. We start by proposing a new technique to improve the quality of the translations generated by fully-automatic machine translation systems. The key insight of our approach is that different translation systems, implementing different approaches and technologies, can exhibit different strengths and limitations. Therefore, a proper combination of the outputs of such different systems has the potential to produce translations of improved quality. We present minimum Bayes¿ risk system combination, an automatic approach that detects the best parts of the candidate translations and combines them to generate a consensus translation that is optimal with respect to a particular performance metric. We thoroughly describe the formalization of our approach as a weighted ensemble of probability distributions and provide efficient algorithms to obtain the optimal consensus translation according to the widespread BLEU score. Empirical results show that the proposed approach is indeed able to generate statistically better translations than the provided candidates. Compared to other state-of-the-art systems combination methods, our approach reports similar performance not requiring any additional data but the candidate translations. Then, we focus our attention on how to improve the utility of automatic translations for the end-user of the system. Since automatic translations are not perfect, a desirable feature of machine translation systems is the ability to predict at run-time the quality of the generated translations. Quality estimation is usually addressed as a regression problem where a quality score is predicted from a set of features that represents the translation. However, although the concept of translation quality is intuitively clear, there is no consensus on which are the features that actually account for it. As a consequence, quality estimation systems for machine translation have to utilize a large number of weak features to predict translation quality. This involves several learning problems related to feature collinearity and ambiguity, and due to the ¿curse¿ of dimensionality. We address these challenges by adopting a two-step training methodology. First, a dimensionality reduction method computes, from the original features, the reduced set of features that better explains translation quality. Then, a prediction model is built from this reduced set to finally predict the quality score. We study various reduction methods previously used in the literature and propose two new ones based on statistical multivariate analysis techniques. More specifically, the proposed dimensionality reduction methods are based on partial least squares regression. The results of a thorough experimentation show that the quality estimation systems estimated following the proposed two-step methodology obtain better prediction accuracy that systems estimated using all the original features. Moreover, one of the proposed dimensionality reduction methods obtained the best prediction accuracy with only a fraction of the original features. This feature reduction ratio is important because it implies a dramatic reduction of the operating times of the quality estimation system. An alternative use of current machine translation systems is to embed them within an interactive editing environment where the system and a human expert collaborate to generate error-free translations. This interactive machine translation approach have shown to reduce supervision effort of the user in comparison to the conventional decoupled post-edition approach. However, interactive machine translation considers the translation system as a passive agent in the interaction process. In other words, the system only suggests translations to the user, who then makes the necessary supervision decisions. As a result, the user is bound to exhaustively supervise every suggested translation. This passive approach ensures error-free translations but it also demands a large amount of supervision effort from the user. Finally, we study different techniques to improve the productivity of current interactive machine translation systems. Specifically, we focus on the development of alternative approaches where the system becomes an active agent in the interaction process. We propose two different active approaches. On the one hand, we describe an active interaction approach where the system informs the user about the reliability of the suggested translations. The hope is that this information may help the user to locate translation errors thus improving the overall translation productivity. We propose different scores to measure translation reliability at the word and sentence levels and study the influence of such information in the productivity of an interactive machine translation system. Empirical results show that the proposed active interaction protocol is able to achieve a large reduction in supervision effort while still generating translations of very high quality. On the other hand, we study an active learning framework for interactive machine translation. In this case, the system is not only able to inform the user of which suggested translations should be supervised, but it is also able to learn from the user-supervised translations to improve its future suggestions. We develop a value-of-information criterion to select which automatic translations undergo user supervision. However, given its high computational complexity, in practice we study different selection strategies that approximate this optimal criterion. Results of a large scale experimentation show that the proposed active learning framework is able to obtain better compromises between the quality of the generated translations and the human effort required to obtain them. Moreover, in comparison to a conventional interactive machine translation system, our proposal obtained translations of twice the quality with the same supervision effort. / González Rubio, J. (2014). On the effective deployment of current machine translation technology [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/37888 / TESIS

Abstracts in English, Xhosa and Southern Sotho / This study critically explores the link between internationalisation theories and outward foreign direct investment (FDI) – a linkage which is well documented in the literature. Numerous studies have established that the internationalisation process recognises both firm- and market-specific aspects, which greatly determine the direction of outward FDI in terms of volume and pattern. In this interaction, path dependency is determined by the intensity of overlapping aspects or linkages, from firm-level heterogeneity and host market aspects that direct investment patterns in terms of the latter, to the volumes of firm-level adjustments. Firm-level heterogeneity comprises those traits, which enable an individual firm to make an investment decision, select a market-entry strategy and create the competitive advantages that will sustain its investments. Macro-level or country-specific aspects encompass those traits or characteristics of host markets, which encourage FDI on the part of multinational enterprises (MNEs). Most studies overlook the path dependency of country- and firm-specific aspects, which are crucial to the internationalisation processes of international business, economics and trade. Academic studies either focus on macro- or micro-level aspects, without paying specific attention to the path dependency of expansion strategies. The present study attempts to fill these gaps in the existing body of knowledge, by investigating international business in these contexts. The rationale for undertaking this study was two-fold: first, FDI holds proven benefits for host markets, which include economic growth, industry spillover, human capital development and transitory tacit knowledge. From a firm-level perspective, outward FDI largely enhances the capacity of MNEs, prompting an increase in asset accumulation, market share and human capital development, the more efficient utilisation of resources and return on equity. In this study, an argument is presented for measuring the variables of both firm- and market-specific aspects, since most existing studies in this genre focus either on micro- or macro-level determinants, or totally overlook the importance of linkages. Second, no documented research has investigated the path dependency of expansion strategies, especially in Africa. Crucially, the importance of path dependency of South Africa’s outward investment has not been documented either. Further, existing evidence on the role the path dependency of expansion strategies plays in outward FDI are scarce, with even fewer studies following a sectorial approach. This study attempts to fill these academic research gaps by reflecting both firm- and market-level data from various sources for the period 1995–2015, using panel dynamic regression models. The study found that the linkages between firm heterogeneity (firm-level evidence) and market-level aspects create a path dependency of expansion strategies. MNEs adopt either joint ventures or wholly owned subsidiaries (or both) as market-entry strategies, but the decision is informed by the intensity of those firm heterogeneity aspects that allow them to exploit opportunities and mitigate risk in host markets. Notably, the intensity of path dependency seemingly varies from one industrial segment to the next. The impulse response approach delivered evidence that one standard deviation shock of firm-specific variables led to a moderate improvement in firm-level capacities in the short run, but a significant improvement in the long run. The same result was recorded for market-level aspects, with the intensity of the results varying from one industry to the next. The causality test attempted to explore the causal relationship between the study variables in both firm- and market-level aspects. Empirical evidence from the study indicates that the size of the firm and its capacity to utilise its resources efficiently, influence their investment in host markets. As regards market-specific aspects, the size of the economy, levels of industry and trade openness were found to have a causal effect on the inflow of FDI in host markets. The intensity of causal aspects was also found to vary from one industry to the next, due to variations in firm-level heterogeneity and their linkage in terms of aspects related to the host market. In sum, this study complements existing material on the subject of international business. / Olu phononongo luphicotha ikhonkco phakathi kweengcingane zamazwe ngamazwe kunye notyalo-mali ngokuthe ngqo oluphuma ngaphandle kumazwe asemzini (i-FDI) –indibaniselo ebhalwe kakuhle kwimiqulu yoncwadi. Izifundo ezininzi ezenziweyo ziye zaqinisekisa ukuba inkqubo yamazwe ngamazwe iyazamkela zombini inkampani- kunye nemiba ekhethekileyo yemarike, ezihlola kakhulukazi imikhombandlela (izikhokelo) ye-FDI yangaphandle ngokomthamo kunye nephatheni. Kule ntsebenziswano, indlela yokuxhomekeka ifunyanwa ngobungakanani bezinto ezisebenzelelanayo/ezingenanayo okanye izenzo zokuhlangana, ukusuka kwiintlobo-ntlobo zamanqanaba enkampani kunye neemfuno zabasingathe imicimbi yeendawo zokuthengisa (iimarike) iimpahla ezilawula iiphatheni zotyalo-imali ngokweyokugqibela, kwimilinganiselo yokulungelelaniswa kwenqanaba lwenkampani. Iintlobo-ntlobo zamanqanaba enkampani ziquka ezo mpawu, ezenza inkampani nganye yenze isigqibo sotyalo-mali, ikhethe isicwangciso sokungeniswa kwimarike kwaye siyile amathuba amahle okhuphiswano aya kugcina utyalo-mali. Inqanaba eliphezulu okanye iinkalo ezithile zelizwe zibandakanya ezo zimo okanye iimpawu zeemarike ezamkelekileyo, ezikhuthaza i-FDI kwiinkampani zamazwe ngamazwe (i-MNEs). Uninzi lwezifundo aziyiniki ngqalelo indlela yokuxhomekeka yelizwe kwimicimbi ekhethekileyo nebalulekileyo yenkampani kwiinkqubo zangokwamazwe oshishino lwamazwe ngamazwe, uqoqosho norhwebo. Uphando lwemfundo ephakamileyo lugxininisa kwiinkcukacha ezikwinqanaba eliphezulu okanye eliphantsi ngokunganiki ngqalelo kwindlela yokuxhomekeka yeendlela zokwandisa. Uphononongo lwangoku luzama ukuvala izikhewu/izikroba kulwazi olukhoyo., ngokuphanda ishishini lwamazwe ngamazwe kule meko. Ingqiqo ekwenzeni olu phando yahlulwe kubini: okokuqala, i-FDI inenzuzo eqinisekisiweyo kwabasingethe iimarike, ezibandakanya ukukhula koqoqosho, ukuchuma kwamashishini, ukuphuhliswa kwezakhono zabantu kunye nolwazi oludlulileyo lwezakhono. Ngakwicala lenqanaba lenkampani, i-FDI yangaphandle iphakamisa amandla e-MNE, ikhawulezisa ukunyusa uqokelelo lwempahla, isabelo semarike kunye nophuhliso lwabantu, ukusetyenziswa ngokufanelekileyo kwezixhobo kunye nokubuyela kubulungisa bokulingana. Kolu phononongo, impikiswano inikezelwe ukulinganisa iinguqu zombini yenkampani- kunye nemimiselo ethile yemarike, njengoko olunye uphando oluninzi olwenziweyo kolu hlobo lugxininisa koonobangela abakwizinga elisezantsi okanye eliphezulu, okanye kunganikwa ngqalelo tu kukubaluleka kwezenzo zokudibana / zokunxibelelana. Okwesibini, akukho phando lubhaliweyo oluphande indlela yokuxhomekeka kweendlela zokwandisa, ngakumbi e-Afrika. Ngokusesikweni, ukubaluleka kwendlela yokuxhomekeka yotyalo-mali lwangaphandle eMzantsi Afrika alukaze nalo lubhalwe phantsi. Ukongezelela, ubungqina obukhoyo kwindima yendlela yokuxhomekeka yeendlela zokwandisa kwi-FDI yangaphandle zinqabile, kwakunye nezifundo ezimbalwa ezilandela indlela yamacandelo. Olu phononongo luzama ukuzalisa izikroba zophando zemfundo ephakamileyo ngokuzibonakalisa zombini inkampani- kunye nedatha yamanqanaba emarike avela kwimithombo eyahlukeneyo yexesha lowe-1995-2015, usebenzisa iimodeli zepaneli ezinamandla zokubuy’umva. Uphononongo lufumanise ukuba ukudibana phakathi kweentlobo-ntlobo zenkampani (ubungqina bezinga lenkampani) kunye nemilinganiselo yezinga lemarike zidala indlela yoxhomekeko yeendlela zokukhula. Ii-MNE zamkela intsebenziswano ngokuhlangeneyo okanye bazibambele ngokwabo ngokupheleleyo (okanye zombini) njengeendlela zokungena kwimarike, kodwa isigqibo siphenjelelwa bubungakanani beentlobo-ntlobo zemicimbi yenkampani evumela ukuba baxhaphaze amathuba kwaye banciphise umngcipheko kwiimarike zenkampani. Ngokuphawulekayo, ubukhulu bokuxhomekeka wendlela yokuxhomekeka kukhangeleka kusahluka ukusuka kwicandelo elinye lozoshishino ukuya kwelinye elilandelayo. Indlela yokuphendula ngokungxama inikezele ubungqina bokuba ukuphazamiseka okusesikweni kwizinto eziguquguqukayo zenkampani ezikhethekileyo zikhokelele ekuphuculeni okusezingeni eliphakathi kwinqanaba kubungakanani benqanaba lenkampani ngexeshana, kodwa ukuphuculwa okubonakalayo nokubalulekileyo ekuhambeni kwexesha. Isiphumo esifanayo sabhalwa phantsi kwiinkalo zemarike, nobukhulu beziphumo zohluka ukusuka kwelinye ishishini ukuya kwelinye. Uvavanyo lwamaxesha athile luzame ukuphonononga ubudlelwane bamaxesha athile phakathi kwezifundo zezinto eziguquguqukayo kwiinkalo zombini inkampani –kunye nenqanaba lemarike/ neemeko zemarike. Ubungqina bamava obuvela kuphando lubonisa ukuba ubungakanani benkampani kunye namandla okusebenzisa uvimba wezixhobo ngokufanelekileyo, ziphembelela utyalo-mali kwiimarike zenkampani. Ngokubhekiselele kwimimandla ethile yemarike, ubungakanani boqoqosho, amazinga oshishino kunye nokuvuleka kwezorhwebo kufunyaniswe ukuba kunefuthe elenzekayo ngamaxesha athile ekungeneni kwe-FDI kubasingathi beemarike. Ubungakanani bemicimbi eyenzeka ngamaxesha athile yafunyanwa kwakhona ukuba yohlukile ukusuka kwelinye ishishini ukuya kwelinye, ngenxa yeenguqu kwiintlobo-ntlobo zeamanqanaba enkampani kunye nokudibana kwabo ngokwemiba enxulumene nabasingethe imarike. Kafutshane esi sifundo, sigcwalisa izixhobo ezikhoyo kwisihloko sezoshishino lamazwe ngamazwe. / Dinyakišišo tše di utolla ka tsinkelo kgokagano gareng ga diteori tša peyomaemong a boditšhabatšhaba le peeletšothwii ye e tšwago dinageng tša ka ntle (FDI) – e lego kgokagano yeo go ngwadilwego ka yona kudu ka dingwalweng. Dinyakišišo tše mmalwa di utollotše gore tshepedišo ya go bea maemong a boditšhabatšhaba e lemoga bobedi dilo tša difeme le tše di amanago le difeme, tšeo di laolago kudu fao FDI ya dinaga tša ka ntle e lebilego gona mabapi le bontši le mokgwa. Ka tirišanong ye, go tšea diphetho go ya ka maemo go laolwa ke bontši bja dilo tšeo di dirwago ka nako e tee goba dikamano, go tloga go go farologanya ditšweletšwa ka femeng le dilo tša mmaraka wa ka nageng tšeo di laolago mekgwa ya dipeeletšo mabapi le go ya ka mmaraka wa ka nageng, go ya go mehuta ye mentši ya dipeakanyo tša ka femeng. Go farologanya ditšweletšwa ka femeng go bopilwe ke diphetogo tše, tšeo di kgontšhago feme ye itšego go tšea sephetho sa mabapi le peeletšo, go kgetha maano a go tsena ka mmarakeng le go hlama menyetla ye mekaone yeo e tlago tšwetša pele peeletšo ya yona. Dikokwane tša ekonomi ye kgolo goba tša ka nageng di akaretša diphetogo tšeo goba dipharologantšhi tša mebaraka ya ka nageng, tšeo di hlohleletšago FDI ka karolong ya dikgwebo tša dinaga tša ka ntle (di-MNE). Dinyakišišo tše ntši di hlokomologile go tšea diphetho go ya ka maemo ga naga le ga dilo tša ka femeng ye itšego, tšeo di lego bohlokwa go tshepedišo ya peyomaemong a boditšhabatšhaba ya dikgwebo tša boditšhabatšhaba, diekonomi le kgwebišano. Dinyakišišo tša dirutegi di ka be di lebeletše kudu dilo tša ekonomi ye kgolo goba tša ye nnyane, ka ntle le go lebiša šedi ye kgolo go go tšea diphetho go ya ka maemo a boditšhabatšhaba ga maano a katološo. Dinyakišišo tše di leka go tlatša dikgoba tše ka tsebo ye e lego gona, ka go nyakišiša dikgwebo tša boditšhabatšhaba ka maemong a. Maikemišetšo a go dira dinyakišišo tše e bile a mabedi: sa mathomo, FDI e na le dikholego tšeo di tiišeditšwego go mebaraka ya ka dinageng, tšeo di akaretšago kgolo ya ekonomi, khuetšano ya diintasteri, tlhabollo ya bokgoni bja bašomi le phetišetšo ya tsebo ye e lego nyanyeng. Go ya ka maemong a difeme, FDI ye e tšwago dinageng tša ka ntle e godiša bokgoni bja di- MNE, ya hlohleletša koketšego ya khwetšo ya dithoto, ya kabelano ya mmaraka le ya tlhabollo ya bokgoni bja bašomi, tšhomišo ye kaone kudu ya methopo le go hwetša poelo go dikabelano. Ka mo dinyakišišong tše, go hlagišwa ntlha ya go ela diphapano tša bobedi dilo tša ka femeng le tša ka mmarakeng, ka ge bontši bja dinyakišišo tše di lego gona ka mo lekaleng le la dinyakišišo di lebeletše kudu tšeo di laolago ekonomi ye nnyane goba ye kgolo goba tšeo di hlokomologago ka botlalo bohlokwa bja dikgokagano. Sa bobedi, ga go dinyakišišo tšeo di ngwadilwego tšeo di nyakišišitšego go tšea diphetho go ya ka maemo ga maano a katološo, kudukudu ka Afrika. Se bohlokwa ke gore, bohlokwa bja go tšea diphetho go ya ka maemo ga peeletšo ya Afrika Borwa ya dinaga tša ka ntle ga se gwa ngwalwa le ge go le bjale. Godimo ga fao, bohlatse bjo bo lego gona ka ga mošomo wa go tšea diphetho go ya ka maemo fao go ralokago ka ga maano a katološo ka go FDI ya dinaga tša ka ntle e se bjo bontši, gomme go na le dinyakišišo tše mmalwa go latela mokgwa wo o lebeletšego makala. Dinyakišišo tše di leka go tlatša dikgoba tše tša dinyakišišo tša dirutegi ka go laetša tshedimošo ya bobedi ka maemong a difeme le ka mebarakeng go tšwa methopong ya mehutahuta go tloga ka mengwaga ya 1995–2015, ka go šomiša mekgwa ya kakanyo ya dikamano ye e fetogago. Dinyakišišo di hweditše gore dikamano gareng ga go farologanya ditšweletšwa (bohlatse bja ka maemong a difeme) le dilo tša maemo a ka mmarakeng di hlola go tšea diphetho go ya ka maemo ga maano a katološo. Di-MNE di šomiša masolo a mohlakanelwa goba dikhamphani tša ka fasana tšeo di laolwago ka botlalo (goba ka bobedi) bjalo ka maano a go tsena ka mmarakeng, eupša sephetho se laolwa ke bontši bja dilo tšeo tša go farologanya ditšweletšwa tšeo di di dumelelago go nyaka dibaka le go fokotša kotsi ka mebarakeng ya ka nageng. Seo se lemogilwego ke gore, bontši bja go tšea diphetho go ya ka maemo go bonala go fapane go ya ka karolo ya intasteri go ya go ye nngwe. Mokgwa wa go arabela kgoketšo wo o hlagišitšwego ka bohlatseng bja gore phapogo ya tlwaelo ya diphapano tša ka femeng e feleleditše ka kaonafalo ya magareng ya bokgoni bja difeme lebakeng le lekopana, eupša ka kaonafalo ye kgolo mo lebakeng le letelele. Dipoelo tše di swanago di begilwe ka go dilo tša maemo a ka mmarakeng, gomme bontši bja dipoelo tša fapana go ya ka diintasteri. Teko ya mathata yeo e bego e leka go utolla kamano ya tšeo di bakago se gareng ga phapano ya dinyakišišo ka go bobedi ka dilo tša ka femeng le tša ka mmarakeng. Bohlatse bja maitemogelo go tšwa ka mo dinyakišišong bo laetša gore bogolo bja feme le bokgoni bja yona bja go šomiša methopo ya yona gabotse ntle le mathata, di huetša peeletšo ya yona ka mebarakeng ya ka nageng. Mabapi le dilo tša ka mmarakeng, bogolo bja ekonomi, maemo a intasteri le go hloka sephiri ka kgwebišanong di bonwe di na le seabe sa go baka seemo go tseneng ga FDI ka mebarakeng ya ka nageng. Bontši bja dilo tše di bakago maemo go hweditšwe gape gore go fapana go ya ka diintasteri, ka lebaka la diphapano ka go farologanyo ya ditšweletšwa ka difemeng le kamano ya tšona mabapi le dilo tšeo di amanago le mmaraka wa ka nageng. Bjalo ka kakaretšo, dinyakišišo tše di tlaleletša dingwalwa tšeo di lego gona ka ga hlogotaba ya dikgwebo tša boditšhabatšhaba. / Business Management / D. Phil. (Business Management)

Dans le cadre de la sélection de modèle par contraste pénalisé, L. Birgé and P. Massart ont prouvé que le phénomène de pénalité minimale se produit pour la sélection libre parmi des variables gaussiennes indépendantes. Nous étendons certains de leurs résultats à la partition d'un signal gaussien lorsque la famille de partitions envisagées est suffisamment riche, notamment dans le cas des arbres de régression. Nous montrons que le même phénomène se produit dans le cadre de l'estimation de densité. La richesse de la famille de modèle s'apparente à une forme d'isotropie. De ce point de vue le phénomène de pénalité minimale est intrinsèque. Pour corroborer et illustrer ce point de vue, nous montrons que le même phénomène se produit pour une famille de modèles d'orientation aléatoire uniforme. / L. Birgé and P. Massart proved that the minimum penalty phenomenon occurs in Gaussian model selection when the model family arises from complete variable selection among independent variables. We extend some of their results to discrete Gaussian signal segmentation when the model family corresponds to a sufficiently rich family of partitions of the signal's support. This is the case of regression trees. We show that the same phenomenon occurs in the context of density estimation. The richness of the model family can be related to a certain form of isotropy. In this respect the minimum penalty phenomenon is intrinsic. To corroborate this point of view, we show that the minimum penalty phenomenon occurs when the models are chosen randomly under an isotropic law.

La presente tesis doctoral surge ante la necesidad creciente por parte de la mayoría de empresas, y en especial (pero no únicamente) aquellas dentro de los sectores farmacéu-tico, químico, alimentación y bioprocesos, de aumentar la flexibilidad en su rango ope-rativo para reducir los costes de fabricación, manteniendo o mejorando la calidad del producto final obtenido. Para ello, esta tesis se centra en la aplicación de los conceptos del Quality by Design para la aplicación y extensión de distintas metodologías ya exis-tentes y el desarrollo de nuevos algoritmos que permitan la implementación de herra-mientas adecuadas para el diseño de experimentos, el análisis multivariante de datos y la optimización de procesos en el ámbito del diseño de mezclas, pero sin limitarse ex-clusivamente a este tipo de problemas. Parte I - Prefacio, donde se presenta un resumen del trabajo de investigación realiza-do y los objetivos principales que pretende abordar y su justificación, así como una introducción a los conceptos más importantes relativos a los temas tratados en partes posteriores de la tesis, tales como el diseño de experimentos o diversas herramientas estadísticas de análisis multivariado. Parte II - Optimización en el diseño de mezclas, donde se lleva a cabo una recapitu-lación de las diversas herramientas existentes para el diseño de experimentos y análisis de datos por medios tradicionales relativos al diseño de mezclas, así como de algunas herramientas basadas en variables latentes, tales como la Regresión en Mínimos Cua-drados Parciales (PLS). En esta parte de la tesis también se propone una extensión del PLS basada en kernels para el análisis de datos de diseños de mezclas, y se hace una comparativa de las distintas metodologías presentadas. Finalmente, se incluye una breve presentación del programa MiDAs, desarrollado con la finalidad de ofrecer a sus usuarios la posibilidad de comparar de forma sencilla diversas metodologías para el diseño de experimentos y análisis de datos para problemas de mezclas. Parte III - Espacio de diseño y optimización a través del espacio latente, donde se aborda el problema fundamental dentro de la filosofía del Quality by Design asociado a la definición del llamado 'espacio de diseño', que comprendería todo el conjunto de posibles combinaciones de condiciones de proceso, materias primas, etc. que garanti-zan la obtención de un producto con la calidad deseada. En esta parte también se trata el problema de la definición del problema de optimización como herramienta para la mejora de la calidad, pero también para la exploración y flexibilización de los procesos productivos, con el objeto de definir un procedimiento eficiente y robusto de optimiza-ción que se adapte a los diversos problemas que exigen recurrir a dicha optimización. Parte IV - Epílogo, donde se presentan las conclusiones finales, la consecución de objetivos y posibles líneas futuras de investigación. En esta parte se incluyen además los anexos. / Aquesta tesi doctoral sorgeix davant la necessitat creixent per part de la majoria d'em-preses, i especialment (però no únicament) d'aquelles dins dels sectors farmacèutic, químic, alimentari i de bioprocessos, d'augmentar la flexibilitat en el seu rang operatiu per tal de reduir els costos de fabricació, mantenint o millorant la qualitat del producte final obtingut. La tesi se centra en l'aplicació dels conceptes del Quality by Design per a l'aplicació i extensió de diferents metodologies ja existents i el desenvolupament de nous algorismes que permeten la implementació d'eines adequades per al disseny d'ex-periments, l'anàlisi multivariada de dades i l'optimització de processos en l'àmbit del disseny de mescles, però sense limitar-se exclusivament a aquest tipus de problemes. Part I- Prefaci, en què es presenta un resum del treball de recerca realitzat i els objec-tius principals que pretén abordar i la seua justificació, així com una introducció als conceptes més importants relatius als temes tractats en parts posteriors de la tesi, com ara el disseny d'experiments o diverses eines estadístiques d'anàlisi multivariada. Part II - Optimització en el disseny de mescles, on es duu a terme una recapitulació de les diverses eines existents per al disseny d'experiments i anàlisi de dades per mit-jans tradicionals relatius al disseny de mescles, així com d'algunes eines basades en variables latents, tals com la Regressió en Mínims Quadrats Parcials (PLS). En aquesta part de la tesi també es proposa una extensió del PLS basada en kernels per a l'anàlisi de dades de dissenys de mescles, i es fa una comparativa de les diferents metodologies presentades. Finalment, s'inclou una breu presentació del programari MiDAs, que ofe-reix la possibilitat als usuaris de comparar de forma senzilla diverses metodologies per al disseny d'experiments i l'anàlisi de dades per a problemes de mescles. Part III- Espai de disseny i optimització a través de l'espai latent, on s'aborda el problema fonamental dins de la filosofia del Quality by Design associat a la definició de l'anomenat 'espai de disseny', que comprendria tot el conjunt de possibles combina-cions de condicions de procés, matèries primeres, etc. que garanteixen l'obtenció d'un producte amb la qualitat desitjada. En aquesta part també es tracta el problema de la definició del problema d'optimització com a eina per a la millora de la qualitat, però també per a l'exploració i flexibilització dels processos productius, amb l'objecte de definir un procediment eficient i robust d'optimització que s'adapti als diversos pro-blemes que exigeixen recórrer a aquesta optimització. Part IV- Epíleg, on es presenten les conclusions finals i la consecució d'objectius i es plantegen possibles línies futures de recerca arran dels resultats de la tesi. En aquesta part s'inclouen a més els annexos. / The present Ph.D. thesis is motivated by the growing need in most companies, and specially (but not solely) those in the pharmaceutical, chemical, food and bioprocess fields, to increase the flexibility in their operating conditions in order to reduce production costs while maintaining or even improving the quality of their products. To this end, this thesis focuses on the application of the concepts of the Quality by Design for the exploitation and development of already existing methodologies, and the development of new algorithms aimed at the proper implementation of tools for the design of experiments, multivariate data analysis and process optimization, specially (but not only) in the context of mixture design. Part I - Preface, where a summary of the research work done, the main goals it aimed at and their justification, are presented. Some of the most relevant concepts related to the developed work in subsequent chapters are also introduced, such as those regarding design of experiments or latent variable-based multivariate data analysis techniques. Part II - Mixture design optimization, in which a review of existing mixture design tools for the design of experiments and data analysis via traditional approaches, as well as some latent variable-based techniques, such as Partial Least Squares (PLS), is provided. A kernel-based extension of PLS for mixture design data analysis is also proposed, and the different available methods are compared to each other. Finally, a brief presentation of the software MiDAs is done. MiDAs has been developed in order to provide users with a tool to easily approach mixture design problems for the construction of Designs of Experiments and data analysis with different methods and compare them. Part III - Design Space and optimization through the latent space, where one of the fundamental issues within the Quality by Design philosophy, the definition of the so-called 'design space' (i.e. the subspace comprised by all possible combinations of process operating conditions, raw materials, etc. that guarantee obtaining a product meeting a required quality standard), is addressed. The problem of properly defining the optimization problem is also tackled, not only as a tool for quality improvement but also when it is to be used for exploration of process flexibilisation purposes, in order to establish an efficient and robust optimization method in accordance with the nature of the different problems that may require such optimization to be resorted to. Part IV - Epilogue, where final conclusions are drawn, future perspectives suggested, and annexes are included. / Palací López, DG. (2018). Quality by Design through multivariate latent structures [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/115489 / TESIS

A new combination of techniques for the numerical computation of incompressible flow is presented. The temporal discretization bases on the discontinuous Galerkin-formulation. Both constant (DG(0)) and linear approximation (DG(1)) in time is discussed. In case of DG(1) an iterative method reduces the problem to a sequence of problems each with the dimension of the DG(0) approach. For the semi-discrete problems a Galerkin/least-squares method is applied. Furthermore a non-overlapping domain decomposition method can be used for a parallelized computation. The main advantage of this approach is the low amount of information which must be exchanged between the subdomains. Due to the slight bandwidth a workstation-cluster is a suitable platform. Otherwise this method is efficient only for a small number of subdomains. The interface condition is of the Robin/Robin-type and for the Navier-Stokes equation a formulation introducing a further pressure interface condition is used. Additionally a suggestion for the implementation of the standard k-epsilon turbulence model with special wall function is done in this context. All the features mentioned above are implemented in a code called ParallelNS. Using this code the verification of this approach was done on a large number of examples ranging from simple advection-diffusion problems to turbulent convection in a closed cavity.

This thesis focuses on detection algorithms of radar targets, namely on group of techniques for removing of disturbing reflections from static objects - clutter and for suppression of distortion products caused by the phase noise of the transmitter and receiver. Methods for distortion suppression in received signal are designed for implementation in the developed active multistatic radar, which operates in the code division multiplex of several transmitters on single frequency. The aim of the doctoral thesis is to design, implement in tool for technical computing MATLAB and analyze the effectiveness and computational complexity of these techniques on simulated and real data.

LIDAR (2D) has been widely used for mapping, localization and navigation in mobile robotics. However, its usage is limited to simple environments. This problem can be solved by adding more sensors and processing these data together. This paper explores a method how measurements from a stereo camera and LIDAR are fused to dynamical mapping. An occupancy grid map from LIDAR data is used as prerequisite and extended by a 2D grid map from stereo camera. This approach is based on the ground plane estimation in disparity map acquired from the stereo vision. For the ground plane detection, RANSAC and Least Squares methods are used. After obstacles determination, 2D occupancy map is generated. The output of this method is 2D map as a fusion of complementary maps from LIDAR and camera. Experimental results obtained from RUDA robot and MIT Stata Center Data Set are good enough to determine that this method is a benefit, although my implementation is still a prototype. In this paper, we present the applied methods, analyze the results and discuss the modifications and possible extensions to get better results.

This diploma thesis deal with analysis of current surveying methods of railway track spatial position of the track section located between Břeclav and Znojmo. At the beginning of the work there is described historical development of railway geodesy in our country and conditions of railway geodetic points field abroad. Further there are mentioned rules for measurement of railway point fields in the Czech Republic. In the following chapter you will find sub-stages of building railway point fields in track section with analysis of accuracy measurement and a proposal of prior accuracy of measurement for computing with the least square method. In the end of the thesis there is described a calculation of parameters spatial track geometry with the suggestion of computing service for uniform way of calculating partial stages of railway geodetic point field.

This thesis focuses on robotic 3D vision for application in Bin Picking. The new method based on Conformal Geometric Algebra (CGA) is proposed and tested for identification of spheres in Pointclouds created with 3D scanner. The speed, precision and scalability of this method is compared to traditional descriptors based method. It is proved that CGA maintains the same precision as the traditional method in much shorter time. The CGA based approach seems promising for the use in the future of robotic 3D vision for identification and localization of spheres.