Full weighted least squares (full WLS) and robust weighted least squares (robust WLS) are currently the two primary estimation methods designed for structural equation modeling with ordinal observed variables. These methods assume that continuous latent variables were coarsely categorized by the measurement process to yield the observed ordinal variables, and that the model proposed by the researcher pertains to these latent variables rather than to their ordinal manifestations. Previous research has strongly suggested that robust WLS is superior to full WLS when models are correctly specified. Given the realities of applied research, it was critical to examine these methods with misspecified models. This Monte Carlo simulation study examined the performance of full and robust WLS for two-factor, eight-indicator confirmatory factor analytic models that were either correctly specified, overspecified, or misspecified in one of two ways. Seven conditions of five-category indicator distribution shape at four sample sizes were simulated. These design factors were completely crossed for a total of 224 cells. Previously findings of the relative superiority of robust WLS with correctly specified models were replicated, and robust WLS was also found to perform better than full WLS given overspecification or misspecification. Robust WLS parameter estimates were usually more accurate for correct and overspecified models, especially at the smaller sample sizes. In the face of misspecification, full WLS better approximated the correct loading values whereas robust estimates better approximated the correct factor correlation. Robust WLS chi-square values discriminated between correct and misspecified models much better than full WLS values at the two smaller sample sizes. For all four model specifications, robust parameter estimates usually showed lower variability and robust standard errors usually showed lower bias. These findings suggest that robust WLS should likely remain the estimator of choice for applied researchers. Additionally, highly leptokurtic distributions should be avoided when possible. It should also be noted that robust WLS performance was arguably adequate at the sample size of 100 when the indicators were not highly leptokurtic. / text

Incomplete data, unknown source-receiver signatures and free-surface reflectivity represent challenges for a successful prediction and subsequent removal of multiples. In this paper, a new method will be represented that tackles these challenges by combining what we know about wavefield (de-)focussing, by weighted convolutions/correlations, and recently developed curvelet-based recovery by sparsity-promoting inversion (CRSI). With this combination, we are able to leverage recent insights from wave physics towards a nonlinear formulation for the multiple-prediction problem that works for incomplete data and without detailed knowledge on the surface effects.

Background and problem: Since the new law took place in 1999, it has been illegal towash a car with substances that can damage the environment on a paved street or on a driveway through a garage. This has conveyed to a new industry where more and more self-service car wash facility have opened around the country. Vetlanda Vägkrog AB has since 2012 been planning to install manual self-service car wash facility at the back of their restaurant business. The authors mission was to make an analysis in order to examine whether an investment of carwashes are lucrative enough for Vetlanda Vägkrog AB. Aim: The study's main objective was to analyze the profitability of an investment in a self-service car wash facility at Vetlanda Vägkrog AB, based on given data. The authors sub-aim was to clarify which factors in general that had played the greatest part in the establishment of a self-service car wash facility. Method: The authors have used an abductive approach in order to fulfill the aim of the study. Furthermore, have the authors used semi-structured interviews in order to gather all empirical data. The interviews were performed on the suppliers, municipal employees and the two owners of Vetlanda Vägkrog AB. The collected data is then explained using theory and henceforth meet the purpose. Conclusion: The results of this study shows that the investment of a self-service car wash facility based on Vetlanda Vägkrog AB conditions is economically efficient and profitable. Net present value method, Pay back and internal rate of return (IRR) is the following methods that the authors consistently have used in order to solve this task. An analysis of three different outcomes were made on the variables that might influence the results, for instance volume and periodic payments has been done in order to get an idea of how sensitive the estimate was. / Bakgrund och problem: Efter den nya lagen som trädde fram 1999 förbjuds tvätt avbilen med ämnen som skadar miljön på en asfalterad gata eller garageuppfart. Detta har medfört till en ny bransch då allt flera självtvätthallar har öppnats runt om i landet. Vetlanda Vägkrog AB har sedan 2012 haft planer på att installera manuella tvätthallar på baksidan av restaurangverksamheten. Vårt uppdrag var att göra en analys där vi granskade om en investering av biltvättar var lukrativt för Vetlanda Vägkrog AB. Syfte: Studiens huvudsyfte var att analysera lönsamheten för en investering i en biltvätthall åt Vetlanda Vägkrog AB, utifrån given data. Delsyftet blev att belysa vilka generella faktorer som hade spelat störst roll vid ett upprättande av en självtvätthall. Metod: För att uppfylla syftet med studien har vi utgått från en abduktiv metod. Vihar genom semistrukturerade intervjuer samlat empiri från leverantörer, kommunalanställda och två av delägarna för Vetlanda Vägkrog AB. Det materialet förklaras sedan med hjälp av teori för att slutligen uppfylla syftet. Slutsats: Resultatet av vår undersökning visar att investering av en självtvätthallutifrån Vetlanda Vägkrog AB förutsättningar är ekonomiskt effektiv och lönsamt. För att lösa uppgiften användes följande metoder payback-metoden, nuvärdemetoden och internräntemetoden.En analys med tre olika utfall gjordes på de variabler som kunde tänkas påverka resultatet, exempelvis volym och särutbetalningar har gjorts för att få en uppfattning påhur känslig kalkylen var. Samtliga utfall påvisade positivt resultat.

Formal language theory, originally developed to model and study our natural spoken languages, is nowadays also put to use in many other fields. These include, but are not limited to, the definition and visualization of programming languages and the examination and verification of algorithms and systems. Formal languages are instrumental in proving the correct behavior of automated systems, e.g., to avoid that a flight guidance system navigates two airplanes too close to each other. This vast field of applications is built upon a very well investigated and coherent theoretical basis. It is the goal of this dissertation to add to this theoretical foundation and to explore ways to make formal languages and their models more expressive. More specifically, we are interested in models that are able to model quantitative features of the behavior of systems. To this end, we define and characterize weighted automata over structures with hierarchical information and over graphs. In particular, we study infinite nested words, operator precedence languages, and finite and infinite graphs. We show Büchi-like results connecting weighted automata and weighted monadic second order (MSO) logic for the respective classes of weighted languages over these structures. As special cases, we obtain Büchi-type equivalence results known from the recent literature for weighted automata and weighted logics on words, trees, pictures, and nested words. Establishing such a general result for graphs has been an open problem for weighted logics for some time. We conjecture that our techniques can be applied to derive similar equivalence results in other contexts like traces, texts, and distributed systems.

Büchi, Elgot and Trakhtenbrot provided a seminal connection between monadic second-order logic and finite automata for both finite and infinite words. This BET- Theorem has been extended by Lautemann, Schwentick and Thérien to context-free languages by introducing a monadic second-order logic with an additional existentially quantified second-order variable. This new variable models the stack of pushdown au- tomata. A fundamental study by Cohen and Gold extended the context-free languages to infinite words. Our first main result is a second-order logic in the sense of Lautemann, Schwentick and Thérien with the same expressive power as ω-context-free languages. For our argument, we investigate Greibach normal forms of ω-context-free grammars as well as a new type of Büchi pushdown automata, the simple pushdown automata. Simple pushdown automata do not use e-transitions and can change the stack only by at most one symbol. We show that simple pushdown automata of infinite words suffice to accept all ω-context-free languages. This enables us to use Büchi-type results recently developed for infinite nested words. In weighted automata theory, many classical results on formal languages have been extended into a quantitative setting. Weighted context-free languages of finite words trace back already to Chomsky and Schützenberger. Their work has been extended to infinite words by Ésik and Kuich. As in the theory of formal grammars, these weighted ω-context-free languages, or ω-algebraic series, can be represented as solutions of mixed ω-algebraic systems of equations and by weighted ω-pushdown automata. In our second main result, we show that (mixed) ω-algebraic systems can be trans- formed into Greibach normal form. We then investigate simple pushdown automata in the weighted setting. Here, we give our third main result. We prove that weighted simple pushdown automata of finite words recognize all weighted context-free languages, i.e., generate all algebraic series. Then, we show that weighted simple ω-pushdown automata generate all ω-algebraic series. This latter result uses the former result together with the Greibach normal form that we developed for ω-algebraic systems. As a fourth main result, we prove that for weighted simple ω-pushdown automata, Büchi-acceptance and Muller-acceptance are expressively equivalent. In our fifth main result, we derive a Nivat-like theorem for weighted simple ω- pushdown automata. This theorem states that the behaviors of our automata are precisely the projections of very simple ω-series restricted to ω-context-free languages. The last result, our sixth main result, is a weighted logic with the same expressive power as weighted simple ω-pushdown automata. To prove the equivalence, we use a similar result for weighted nested ω-word automata and apply our present result of expressive equivalence of Muller and Büchi acceptance.

In 1994, Alur and Dill introduced timed automata as a simple mathematical model for modelling the behaviour of real-time systems. In this thesis, we extend timed automata with weights. More detailed, we equip both the states and transitions of a timed automaton with weights taken from an appropriate mathematical structure. The weight of a transition determines the weight for taking this transition, and the weight of a state determines the weight for letting time elapse in this state. Since the weight for staying in a state depends on time, this model, called weighted timed automata, has many interesting applications, for instance, in operations research and scheduling. We give characterizations for the behaviours of weighted timed automata in terms of rational expressions and logical formulas. These formalisms are useful for the specification of real-time systems with continuous resource consumption. We further investigate the relation between the behaviours of weighted timed automata and timed automata. Finally, we present important decidability results for weighted timed automata.

In cluster analysis, a fundamental problem is to determine the best estimate of the number of clusters, which has a deterministic effect on the clustering results. However, a limitation in current applications is that no convincingly acceptable solution to the best-number-of-clusters problem is available due to high complexity of real data sets. In this dissertation, we tackle this problem of estimating the number of clusters, which is particularly oriented at processing very complicated data which may contain multiple types of cluster structure. Two new methods of choosing the number of clusters are proposed which have been shown empirically to be highly effective given clear and distinct cluster structure in a data set. In addition, we propose a sequential type of clustering approach, called multi-layer clustering, by combining these two methods. Multi-layer clustering not only functions as an efficient method of estimating the number of clusters, but also, by superimposing a sequential idea, improves the flexibility and effectiveness of any arbitrary existing one-layer clustering method. Empirical studies have shown that multi-layer clustering has higher efficiency than one layer clustering approaches, especially in detecting clusters in complicated data sets. The multi-layer clustering approach has been successfully implemented in clustering the WTCHP microarray data and the results can be interpreted very well based on known biological knowledge. Choosing an appropriate clustering method is another critical step in clustering. K-means clustering is one of the most popular clustering techniques used in practice. However, the k-means method tends to generate clusters containing a nearly equal number of objects, which is referred to as the ``equal-size'' problem. We propose a clustering method which competes with the k-means method. Our newly defined method is aimed at overcoming the so-called ``equal-size'' problem associated with the k-means method, while maintaining its advantage of computational simplicity. Advantages of the proposed method over k-means clustering have been demonstrated empirically using simulated data with low dimensionality. / Ph. D.

The objective of this thesis was to investigate how the addition of a privacy-preserving mechanism to a federated learning model was affecting the performance of the model for an image classification task. Further, it was to get knowledge on how the outlook to use federated learning in the biotech industry is and what possible threats and attacks that could obstruct the utilization of federated learning among competitors. In the project four different aggregation algorithms for federated learning were examined. The methods were weighted fedAvg, unweighted FedAvg, weighted FedProx and unweighted FedProx. The experiment was using tensorflow federated to simulate the different methods. They were evaluated using accuracy, loss, recall, precision and F1 score. The result of this study shows that the performance of the deep neural network model is decreasing as differential privacy is introduced to the process. Out of the four aggregation algorithms used, weighted fedProx was the one that performed the best despite the added noise. It was also concluded that federated learning has potential to be used in the biotechnology industry among competitors, but that there are still security threats and attacks to avoid.

No description available.

Après traitement d’un AIC par thrombolyse, les lésions ischémiques en diffusion peuvent être réversibles, stables et/ou progresser. Cette thèse a pour objectif d’étudier en IRM ces phénomènes de réversibilité et de progression, de mieux comprendre leur physiopathologie, et d’appréhender le rôle pronostic du volume des lésions en diffusion. La réversibilité en diffusion a été rapportée après recanalisation artérielle, chez l’animal et chez l’homme. Dans notre cohorte de 155 patients, nous avons étudié ce phénomène par une analyse voxel à voxel sur des IRM avant et 24h après thrombolyse. Nous avons montré que la réversibilité des anomalies en diffusion à 24h était fréquente et surtout persistait sur une IRM plus tardive (médiane 54h) pour plus de 70% des voxels [58.0-85.9], suggérant que ce phénomène n’était pas transitoire. De plus, seule la réversibilité permanente était associée à une amélioration neurologique à 24h (OR=1.15, CI95%[ 1.03-1.27], P=0.008 par mL). La substance blanche a été décrite comme plus résistante à l’ischémie sur des modèles murins. Chez l’homme, nous avons confirmé la prédominance en substance blanche des phénomènes de réversibilité, avec une probabilité de régression plus importante pour un voxel de substance blanche que pour un voxel de substance grise. Ces résultats pourraient aider à identifier avant décision thérapeutique les lésions potentiellement réversibles, c'est-à-dire celles prédominant en substance blanche. Même s’il existe une association entre le volume de lésion en diffusion avant traitement et le pronostic fonctionnel à 3 mois, la réversibilité des lésions pourrait remettre en cause les seuils de volume au-delà desquels une revascularisation pourrait être futile. En effet, en limitant la progression des anomalies en diffusion et en favorisant leur régression, la revascularisation pourrait aussi avoir un impact sur le pronostic fonctionnel des patients avec un large volume lésionnel (≥70mL). Dans une population de 267 patients avec un infarctus de l’artère cérébrale moyenne traités par thrombolyse, 54 avaient un volume en diffusion ≥70mL, dont 12(22%) avaient un pronostic fonctionnel favorable à 3 mois. L’odds-ratio de la recanalisation pour le pronostic fonctionnel favorable dans le groupe ≥70mL était de 4.87 [1.15-20.73], P=0.03 en faveur d’un impact positif de la recanalisation même dans ce groupe. En l’absence de reperfusion, les anomalies en diffusion progressent au sein de la pénombre ischémique. Selon le modèle « core/pénombre », cette progression lésionnelle ne doit pas s’accompagner d’une aggravation neurologique. Nous avons formulé l’hypothèse que si elle survenait au-delà de la pénombre (dans des zones asymptomatiques), elle s’accompagnerait d’une détérioration neurologique. La détérioration neurologique précoce non expliquée (absence de transformation hémorragique, d’œdème malin ou d’autre cause identifiable) est observée chez 7% de nos 309 patients thrombolysés. Nous avons validé l’hypothèse initiale d’une progression des anomalies en diffusion au-delà de la pénombre chez ces patients. Par une analyse voxel à voxel, elle était présente chez 9 cas/10 (7-137mL; > 10mL chez 8 cas), et son volume était significativement supérieur à celui des 30 contrôles sans détérioration précoce (P=0.047). De plus, sa topographie était congruente avec les fonctions neurologiques qui s’étaient détériorées. L’ensemble de ces résultats participe à une meilleure compréhension de l’évolution des lésions ischémiques aiguës en IRM de diffusion et offre des perspectives pour adapter la prise en charge individuelle afin d’améliorer le pronostic fonctionnel. / Within the first 24 hours after IV-rtPA, diffusion-weighted ischemic lesions can reverse, remain stable or grow. The aim of this thesis is to study these reversal and growth phenomena, to better understand their pathophysiology and to gain insight into the prognostic value of diffusion lesion volume. Diffusion lesion reversal was reported in animals and humans after arterial recanalization. In our sample of 155 patients, we studied this phenomenon with MRI performed before and 24 hours after thrombolysis using a voxel-based approach. First, we demonstrated that 24 hours diffusion reversal was frequent and sustained on a late MRI (median 54h) for over two-thirds of the voxels. Second, sustained reversal was associated with 24hr neurological improvement (OR=1.15, IC95%[ 1.03-1.27], P=0.008 per 1mL). Animal studies have suggested that white matter is more resistant to ischemia than gray matter. In humans, we confirmed that diffusion lesion reversal was more frequent in white matter than in gray matter and disclosed that white matter voxels were more prone to reverse than gray matter voxels. The amount of white matter in the initial diffusion lesion may therefore be a significant determinant of reversibility. Large diffusion lesion volume is associated with poor outcome. However, revascularization therapy can prevent infarct growth or even promote lesion reversal. It is still unclear whether these treatments are beneficial in patients with large diffusion volumes (≥70mL). In our series including 267 patients with middle cerebral artery stroke treated with thrombolysis, 54 patients had a ≥70mL diffusion volume, of which 12(22%) had a 3 month favorable outcome. Odds-ratio of recanalization for favorable outcome in the ≥70mL group was 4.87 [1.15-20.73], P=0.03 supporting a benefit of recanalization in this subgroup. Diffusion lesions growth is usually located within the ischemic penumbra. We hypothesized that if it occurred beyond its boundaries, it would translate into neurological deterioration. Unexplained early neurological deterioration is frequent (70% of early neurological deteriorations, 7% in our series of 309 thrombolysed patients) and its causes are not well-known (no symptomatic intra cerebral hemorrhage, malignant edema or post-stroke seizure). We confirmed our hypothesis in these patients with diffusion growth beyond the penumbra. This growth occurred in 9 of the 10 studied patients (7-137mL; > 10mL in 8 patients) and it was significantly larger than in the 30 controls with early neurological deterioration (P=0.047). Moreover its topography matched the neurological items that deteriorated. All together, these results contribute to a better understanding of acute ischemic lesions using diffusion-weighted imaging and may offer perspectives to adapt individual patient care.