Multiplex networks describe a set of entities, with multiple relationships among them, as a collection of networks over a common vertex set. Multiplex networks naturally describe complex systems where units connect across different modalities whereas single network data only permits a single relationship type. Joint spectral embedding methods facilitate analysis of multiplex network data by simultaneously mapping vertices in each network to points in Euclidean space, entitled node embeddings, where statistical inference is then performed. This mapping is performed by spectrally decomposing a matrix that summarizes the multiplex network. Different methods decompose different matrices and hence yield different node embeddings. This dissertation analyzes a class of joint spectral embedding methods which provides a foundation to compare these different approaches to multiple network inference. We compare joint spectral embedding methods in three ways. First, we extend the Random Dot Product Graph model to multiplex network data and establish the statistical properties of node embeddings produced by each method under this model. This analysis facilitates a full bias-variance analysis of each method and uncovers connections between these methods and methods for dimensionality reduction. Second, we compare the accuracy of algorithms which utilize these different node embeddings in a variety of multiple network inference tasks including community detection, vertex anomaly detection, and graph hypothesis testing. Finally, we perform a time and space complexity analysis of each method and present a case study in which we analyze interactions between New England sports fans on the social news aggregation and discussion website, Reddit. These findings provide a theoretical and practical guide to compare joint spectral embedding techniques and highlight the benefits and drawbacks of utilizing each method in practice.

No description available.

This study examines the ways early leaders can influence the process of institutionalization in public organizations. Using Schein's (1983, 1991) model of cultural creation and embedding as a heuristic device, secondary historical sources detailing the creation and development of the Central Intelligence Agency (CIA) and the Federal Bureau of Investigation (FBI) and the careers of three significant leaders are used to understand the institutionalizing effects of those leaders, how they created those effects, and what happened to those effects over time. The case studies of William Donovan and Allen Dulles at CIA and J. Edgar Hoover at the FBI, provide evidence that these early leaders explicitly and implicitly used several of the cultural creation and embedding mechanisms identified by Schein to entrench their beliefs and predispositions into their organizations. These ensconced attitudes and tendencies seemingly played significant roles in the institutionalization of beliefs, rules, and roles that have developed, persisted, and affected the historical evolution of both CIA and the FBI. / Ph. D.

The Collaborative for Academic, Social, and Emotional Learning (CASEL) (2018) specified that "integrating SEL (Social-Emotional Learning) with instructional practices and academic content has become a growing priority" (p. 1). This priority originates from research that suggests SEL promotes positive student and long-term community outcomes, particularly in secondary schools. This canon of research, however, only reviews the outcomes of implementing purchasable curricula, not content-embedded SEL. The effectiveness of content-embedded SEL instruction, which comprises a large portion of how SEL is implemented at the secondary level (CASEL, 2018; Hart et al., 2013), cannot be effectively measured or predicted because there is little to no identified research regarding three essential factors: how teachers embed SEL, how teachers assess content-embedded SEL, and teachers' perceived readiness to embed SEL. The purpose of this research was to identify the methods secondary teachers indicate they use to implement and assess content-embedded SEL instruction and their perceived preparedness to do so. Educational leaders may be better able to evaluate the effectiveness of content-embedded SEL instruction and improve its implementation with this knowledge. Using a qualitative design, secondary teachers were interviewed to identify how they embed SEL into their instruction, how they assess SEL, and how prepared they perceive they are to deliver content-embedded SEL instruction. This research suggests that expectations for embedding and documenting SEL vary, but teachers appear to be implementing content-embedded SEL nevertheless. Furthermore, teachers recognize that pre-curated resources or lessons are provided to assist them in embedding SEL but appear to rely heavily upon their own teacher-created resources. Among these activities, teachers rely upon opportunities for reflection and choice and voice activities, but no one instructional strategy or manipulative was preferred overall. Teachers perceive student progress in SEL via observation of student behaviors, interactions, and responses both formally and informally. Regarding their preparedness to teach SEL, teachers perceive that their personal SEL proficiencies directly affect their abilities to teach them. Finally, teachers prefer experiential professional learning situations for SEL, and perceive that time to revisit and reflect in smaller, collaborative settings is an effective process for learning to implement SEL, including the use of specialists. / Doctor of Education / Social-Emotional Learning (SEL) is an improvement strategy that has gained popularity in the past decade. Results from research that suggest SEL develops beneficial student and long-term community effects have led to substantial efforts to spread SEL instruction, especially in middle and high schools. However, the research upon which these efforts are based only reviews the benefits of using purchasable programming, not SEL that teachers embed into their content. The success of content-embedded SEL, which makes up a large percentage of how SEL is employed in middle and high schools (CASEL, 2018; Hart et al., 2013), cannot be accurately measured or predicted because there is little to no identified information about three important factors: how teachers embed SEL, how teachers measure content-embedded SEL, and teachers' perceived readiness to embed SEL. The purpose of this research was to identify the methods middle and high school teachers indicate they use to embed and measure SEL and their perceived preparedness to do so. Educational leaders may be better able to measure the success of content-embedded SEL and improve its use with this knowledge. Middle and high school teachers were interviewed to identify how they embed SEL, how they measure SEL, and how prepared they perceive they are to embed SEL. This research suggests that expectations for embedding and recording SEL vary, but teachers still appear to be embedding SEL. Additionally, teachers understand that pre-curated resources or lessons are provided to assist them in embedding SEL but appear to rely more heavily upon their own resources. Among these activities, teachers rely upon opportunities for reflection and choice and voice activities, but no one teaching strategy was preferred overall. Teachers recognize student development in SEL via observation of their behaviors, interactions, and responses; they grade this development about half of the time. Teachers believe their personal SEL proficiencies directly affect their abilities to teach them. Finally, teachers prefer hands-on situations for learning how to embed SEL, and perceive that time to revisit and reflect in smaller, collaborative settings to be an effective process for learning to implement SEL, including the use of specialists.

Short-term load forecasting (STLF) deals with forecasting of hourly system demand with a lead time ranging from one hour to 168 hours. The basic objective of the STLF is to provide for economic, reliable and secure operation of the power system. This dissertation establishes a new approach to artificial neural network (ANN) based STLF. It first decomposes the prediction problem into representation and function approximation problems. The representation problem is solved using phase-space embedding which identifies time delay variables from load time series that are used in forecasting. The concept is inherently different from the methods used so far because it does not use correlated variables for forecasting. Temperature variables are included as well using identified embedding parameters. Function approximation problem is approached using ANN ensemble and active selection of a training set. Training set is selected based on predicted weather parameters for a prediction horizon. Selection is done applying the k-nearest neighbors technique in a temperature-based vector space. A novel approach of pilot set simulation is used to determine the number of hidden units for every forecast period. Ensemble consists of two ANNs which are trained and cross validated on complementary training sets. Final prediction is obtained by a simple average of two trained ANNs. The described technique is used for predicting one week’s load in four selected months in summer peaking and winter peaking US utilities. Mean absolute percent errors (MAPEs) for 24-hour lead time predictions are slightly greater than 2% for all months. For 120-hour lead time (weekday) predictions, MAPEs are around 2.3%. MAPEs for 48- hour lead time (weekend) predictions are around 2.5%. Maximal errors for these cases are around 7%. Predictions for one-hour lead time are slightly higher than 1% for all months, with maximal errors not exceeding 4.99%. Peak load MAPEs are 2.3% for both utilities. Maximal peak-load errors do not exceed 6%. The technique shows very good performance faced with sudden and large changes in weather. For changes in temperature larger than 20° F for two consecutive days, forecasting error is smaller than 3.58%. / Ph. D.

Yes / This paper reports on a regionally based UK study uncovering what has worked well in learning from adverse incidents in hospitals. The purpose of this paper is to review the incident investigation methodology used in identifying strengths or weaknesses and explore the use of a database as a tool to embed learning. Documentary examination was conducted of all adverse incidents reported between 1 June 2011 and 30 June 2012 by three UK National Health Service hospitals. One root cause analysis report per adverse incident for each individual hospital was sent to an advisory group for a review. Using terms of reference supplied, the advisory group feedback was analysed using an inductive thematic approach. The emergent themes led to the generation of questions which informed seven in-depth semi-structured interviews. “Time” and “work pressures” were identified as barriers to using adverse incident investigations as tools for quality enhancement. Methodologically, a weakness in approach was that no criteria influenced the techniques which were used in investigating adverse incidents. Regarding the sharing of learning, the use of a database as a tool to embed learning across the region was not supported. Softer intelligence from adverse incident investigations could be usefully shared between hospitals through a regional forum. The use of a database as a tool to facilitate the sharing of learning from adverse incidents across the health economy is not supported.

The size and conformational complexity of proteins and other large systems represent major challenges for today's methods of quantum chemistry.This thesis is centered around the development of new computational tools to gain molecular-level insight into electronic transitions in such systems. To meet this challenge, we focus on the polarizable embedding (PE) model, which takes advantage of the fact that many electronic transitions are localized to a smaller part of the entire system.This motivates a partitioning of the large system into two regions that are treated at different levels of theory:The smaller part directly involved in the electronic process is described using accurate quantum-chemical methods, while the effects of the rest of the system, the environment, are incorporated into the Hamiltonian of the quantum region in an effective manner. This thesis presents extensions of the PE model with theaim of expanding its range of applicability to describe electronic transitions in large molecular systemsin the optical and X-ray regions. The developments cover both improvements with regardto the quantum region as well as the embedding potential representing the environment.Regarding the former, a damped linear response formulation has been implemented to allow for calculations of absorption spectra of large molecular systems acrossthe entire frequency range. A special feature of this development is its abilityto address core excitations that are otherwise not easily accessible.Another important development presented in this thesis is the coupling of the PE model to a multi-configuration self-consistent-field description of the quantum region and its further combination with response theory. In essence, this extends the PE model to the study of electronic transitions in large systems that are prone to static correlation --- a situation that is frequently encountered in biological systems. In addition to the direct environmental effects on the electronic structure of the quantum region, another important component of the description of electronic transitions in large molecular systems is an accurate account of the indirect effects of the environment, i.e., the geometrical distortions in the quantum region imposed by the environment. In thisthesis we have taken the first step toward the inclusion of geometry distortions in the PE frameworkby formulating and implementing molecular gradients for the quantum region. To identify critical points related to the environment description, we perform a theoretical analysis of the PE model starting from a full quantum-mechanicaltreatment of a composite system. Based on this, we present strategies for an accurate yet efficient construction of the embedding potentialcovering both the calculation of ground state and transition properties. The accurate representation of the environment makes it possible to reduce the size of the quantum region without compromising the overall accuracy of the final results. This further enables use of highly accurate quantum-chemical methods despite their unfavorable scaling with the size of the system. Finally, some examples of applications will be presented to demonstrate how the PE model may be applied as a tool to gain insight into and rationalize the factors influencing electronic transitions in large molecular systems of increasing complexity. / <p>The dissertation was awarded the best PhD thesis prize 2016 by the Danish Academy of Natural Sciences.</p><p></p><p>QC 20170209</p>

L'assistance informatique est devenue une partie indispensable pour la réalisation de procédures chirurgicales modernes. Le désir de créer une nouvelle génération de blocs opératoires intelligents a incité les chercheurs à explorer les problèmes de perception et de compréhension automatique de la situation chirurgicale. Dans ce contexte de prise de conscience de la situation, un domaine de recherche en plein essor adresse la reconnaissance automatique du flux chirurgical. De grands progrès ont été réalisés pour la reconnaissance des phases et des gestes chirurgicaux. Pourtant, il existe encore un vide entre ces deux niveaux de granularité dans la hiérarchie du processus chirurgical. Très peu de recherche se concentre sur les activités chirurgicales portant des informations sémantiques vitales pour la compréhension de la situation. Deux facteurs importants entravent la progression. Tout d'abord, la reconnaissance et la prédiction automatique des activités chirurgicales sont des tâches très difficiles en raison de la courte durée d'une activité, de leur grand nombre et d'un flux de travail très complexe et une large variabilité. Deuxièmement, une quantité très limitée de données cliniques ne fournit pas suffisamment d'informations pour un apprentissage réussi et une reconnaissance précise. À notre avis, avant de reconnaître les activités chirurgicales, une analyse soigneuse des éléments qui composent l'activité est nécessaire pour choisir les bons signaux et les capteurs qui faciliteront la reconnaissance. Nous avons utilisé une approche d'apprentissage profond pour évaluer l'impact de différents éléments sémantiques de l'activité sur sa reconnaissance. Grâce à une étude approfondie, nous avons déterminé un ensemble minimum d'éléments suffisants pour une reconnaissance précise. Les informations sur la structure anatomique et l'instrument chirurgical sont de première importance. Nous avons également abordé le problème de la carence en matière de données en proposant des méthodes de transfert de connaissances à partir d'autres domaines ou chirurgies. Les méthodes de ''word embedding'' et d'apprentissage par transfert ont été proposées. Ils ont démontré leur efficacité sur la tâche de prédiction d'activité suivante offrant une augmentation de précision de 22%. De plus, des observations pertinentes / Computer assistance became indispensable part of modern surgical procedures. Desire of creating new generation of intelligent operating rooms incited researchers to explore problems of automatic perception and understanding of surgical situations. Situation awareness includes automatic recognition of surgical workflow. A great progress was achieved in recognition of surgical phases and gestures. Yet, there is still a blank between these two granularity levels in the hierarchy of surgical process. Very few research is focused on surgical activities carrying important semantic information vital for situation understanding. Two important factors impede the progress. First, automatic recognition and prediction of surgical activities is a highly challenging task due to short duration of activities, their great number and a very complex workflow with multitude of possible execution and sequencing ways. Secondly, very limited amount of clinical data provides not enough information for successful learning and accurate recognition. In our opinion, before recognizing surgical activities a careful analysis of elements that compose activity is necessary in order to chose right signals and sensors that will facilitate recognition. We used a deep learning approach to assess the impact of different semantic elements of activity on its recognition. Through an in-depth study we determined a minimal set of elements sufficient for an accurate recognition. Information about operated anatomical structure and surgical instrument was shown to be the most important. We also addressed the problem of data deficiency proposing methods for transfer of knowledge from other domains or surgeries. The methods of word embedding and transfer learning were proposed. They demonstrated their effectiveness on the task of next activity prediction offering 22% increase in accuracy. In addition, pertinent observations about the surgical practice were made during the study. In this work, we also addressed the problem of insufficient and improper validation of recognition methods. We proposed new validation metrics and approaches for assessing the performance that connect methods to targeted applications and better characterize capacities of the method. The work described in this these aims at clearing obstacles blocking the progress of the domain and proposes a new perspective on the problem of surgical workflow recognition.

Nous avons réalisé une déformation non-triviale et complète de la théorie de supergravité maximale en dimension deux. Il s'agit de la supergravité maximale avec groupe de jauge SO(9). Cette théorie décrit de manière effective la supergravité de type IIA sur un espace-temps produit AdS_2 x S^8. Elle joue ainsi un rôle important dans la correspondance Gravité / Théorie de Jauge appliquée au cas de la D0-brane. Afin de préparer la construction de la supergravité maximale jaugée SO(9), nous nous intéressons aux supergravités maximales en dimension onze et trois, puisqu'elles donnent lieu à différentes formulations non équivalentes de la théorie bidimensionnelle non jaugée. Le formalisme d' « Embedding tensor » est ensuite présenté. Il permet de déterminer l'ensemble des groupes de jauges compatibles avec la supersymétrie maximale. La supergravité SO(9) est dès lors explicitement construite et ouvre la voie à deux applications importantes. P our commencer, nous avons réalisé l'inclusion d'un sous-secteur bosonique de la théorie SO(9), la troncature de Cartan, dans la supergravité de type IIA à dix dimensions d'espace-temps. Il s'agit d'une inclusion cohérente. Cela a motivé la deuxième application, de nature holographique. Ainsi, à partir du sous-secteur de Cartan de la supergravité SO(9), et en particulier de la découverte d'états fondamentaux de type « half-BPS », nous avons calculé un ensemble de fonctions de corrélation à un et deux points associées à des opérateurs de modèles de matrice duaux. Nous avons conclu en un résumé de nos travaux et en la présentation d'intéressantes perspectives. / A complete non trivial supersymmetric deformation of the maximal supergravity in two dimensions is achieved by the gauging of a SO(9) group. The resulting theory describes the reduction of type IIA supergravity on an AdS_2 x S^8 background and is of first importance in the Domain-Wall / Quantum Field theory correspondence for the D0-brane case. To prepare the construction of the SO(9) gauged maximal supergravity, we focus on the eleven dimensional supergravity and the maximal supergravity in three dimensions since they give rise to important off-shell inequivalent formulations of the ungauged theory in two dimensions. The embedding tensor formalism is presented, allowing for a general desciption of the gaugings consistent with supersymmetry. The SO(9) supergravity is explicitly constructed and applications are considered. In particular, an embedding of the bosonic sector of the two-dimensional theory into type IIA supergravity is obtained. Hence, the Cartan truncation of the SO(9) supergravity is proved to be consistent. This motivated holographic applications. Therefore, correlation functions for operators in dual Matrix models are derived from the study of gravity side excitations around half BPS backgrounds. These results are fully discussed and outlooks are presented.

This thesis deals with a complex analysis and prediction of foreign exchange markets. It uses advanced artificial intelligence methods, namely neural networks and chaos theory. It introduces unconventional approaches and methods of each of these areas, compares them and uses on a real problem. The core of this thesis is a comparison of several prediction models based on completely different principles and underlying theories. The outcome is then a selection of the most appropriate prediction model called NAR + H. The model is evaluated according to several criteria, the pros and cons are discussed and approximate expected profitability and risk are calculated. All analytical, prediction and partial algorithms are implemented in Matlab development environment and form a unified library of all used functions and scripts. It also may be considered as a secondary main outcome of the thesis.