Applying machine learning methods to avalanche forecasting

Avalanche forecasting is a complex process involving the assimilation of multiple data sources to make predictions over varying spatial and temporal resolutions. Numerically assisted forecasting often uses nearest neighbour methods (NN), which are known to have limitations when dealing with high dimensional data. We apply Support Vector Machines to a dataset from Lochaber, Scotland to assess their applicability in avalanche forecasting. Support Vector Machines (SVMs) belong to a family of theoretically based techniques from machine learning and are designed to deal with high dimensional data. Initial experiments showed that SVMs gave results which were comparable with NN for categorical and probabilistic forecasts. Experiments utilising the ability of SVMs to deal with high dimensionality in producing a spatial forecast show promise, but require further work.

Multi-sensor Simultaneous Localization and Mapping

Simultaneous localization and mapping(SLAM) is a very important problem in mobile robotics. Many solutions have been proposed by different scientists during the last two decades, nevertheless few studies have considered the use of multiple sensors simultane¬ously. The solution is on combining several data sources with the aid of an Extended Kalman Filter (EKF). Two approaches are proposed. The first one is to use the ordinary EKF SLAM algorithm for each data source separately in parallel and then at the end of each step, fuse the results into one solution. Another proposed approach is the use of multiple data sources simultaneously in a single filter. The comparison of the computational com¬plexity of the two methods is also presented. The first method is almost four times faster than the second one.

Current State of Ontology Matching. A Survey of Ontology and Schema Matching

Ontology matching is an important task when data from multiple data sources is integrated. Problems of ontology matching have been studied widely in the researchliterature and many different solutions and approaches have been proposed alsoin commercial software tools. In this survey, well-known approaches of ontologymatching, and its subtype schema matching, are reviewed and compared. The aimof this report is to summarize the knowledge about the state-of-the-art solutionsfrom the research literature, discuss how the methods work on different application domains, and analyze pros and cons of different open source and academic tools inthe commercial world.

Les relations entre le transport actif et l'environnement urbain : le cas d'usagers du train de banlieue de Montréal

Dans une société où il est plutôt normal de passer beaucoup de temps assis, nous étudions, à partir de l’aménagement, l’intégration de l’activité physique de loisirs et de transport dans les activités et les lieux du quotidien. Cette intégration est relativement peu étudiée dans sa globalité car elle nécessite de prendre en considération les facteurs de l’environnement physique et social, les deux types d’activité physique, les différents lieux fréquentés quotidiennement et elle pose en ce sens de nombreux défis d’ordre méthodologique. Cette vue globale du phénomène s’impose car de plus en plus de recherches font état d’associations entre des dimensions spécifiques de l’aménagement et des comportements précis; occasionnellement, ces résultats se contredisent. Pour comprendre le phénomène, nous sommes partis du modèle écoenvironnemental et l’avons adapté pour mieux représenter la mobilité de la population. Nous avons conséquemment choisi une unité d’analyse comprenant le territoire résidentiel, le territoire du milieu de travail et le trajet entre les deux. Ainsi, en utilisant plusieurs sources de données, nous avons caractérisé des milieux comme étant contraignants ou facilitants pour l’activité physique et les personnes y résidant comme étant suffisamment actives ou pas. Nous avons ensuite fait ressortir les éléments importants des entrevues en fonction de cet appariement. Parmi les thèmes explorés en entrevue, nommons les caractéristiques de l’environnement physique qui ont de l’importance, l’impact de l’environnement social au travail et au domicile, la logique sous-jacente aux courses, etc. Les principaux résultats de cette recherche démontrent que les usagers du train de banlieue font suffisamment d’activité physique en dépit qu’ils résident en banlieue. En ce sens, notre échantillon est plus actif que la moyenne québécoise. Nous remarquons que l’influence de l’environnement est manifeste mais sous le principe des vases communicants, c'est-à-dire que le pôle résidentiel et le pôle des emplois ont tous deux des contributions qui s’avèrent très souvent complémentaires. L’influence de l’environnement social passe par le rôle signifiant des proches plutôt que par leur proximité géographique tandis que l’aménagement a une énorme contribution à rendre les parcours agréables et, de ce fait, donner une plus-value au temps requis pour les emprunter. La vocation des milieux, le type de marche et le sens qu’y voient les usagers doivent guider le design; il n’y a donc pas qu’une formule ou une seule prescription pour augmenter le potentiel piétonnier et/ou cyclable des milieux. Cela dit, les outils de caractérisation doivent être revus. En conclusion des pistes de développements futurs à cette recherche sont proposées.

Analyse de l’implantation d’un plan d’action pour le renforcement du rôle professionnel de la sage-femme dans le Royaume du Maroc

Dans le cadre d’une stratégie nationale visant les objectifs du Millénaire pour le développement 4 et 5 au Maroc - réduire la mortalité maternelle et infantile -, un plan d’action a été développé au sein des trois systèmes (socioculturel, éducationnel, disciplinaire) dans lesquels évolue un rôle professionnel de la santé et ce, pour renforcer le rôle professionnel de la sage-femme. La présente thèse vise à évaluer le niveau d’implantation du plan d’action et à comprendre les facteurs contextuels ayant affecté son implantation et susceptibles d’empêcher l’atteinte de ses effets. Le cadre conceptuel adopté dérive du modèle de Hatem-Asmar (1997) concernant l’interaction entre les systèmes éducationnel, disciplinaire et socioculturel pour changer un rôle professionnel de la santé; et le cadre de Damschroder et al. (2009) pour l’analyse de l’implantation d’une intervention en santé. Le devis est une étude de cas unique à trois niveaux d’analyse. Les données sont recueillies à partir de multiples sources de données : 11 entrevues individuelles semi-structurées, 20 groupes de discussion, observations d’activités de formation, analyse de documents. Les résultats ont montré des déficits notables au niveau de l’implantation. Seize barrières et sept facilitateurs ont été catégorisés sous les construits du cadre de Damschroder et al. (2009) et sous les dimensions des trois systèmes. Un alignement inadéquat entre les dimensions (valeurs, méthodes, acteurs et finalités) du système socioculturel et celles (valeurs, méthodes, acteurs) des systèmes éducationnel et disciplinaire d’une part, avec le plan d’action d’autre part empêche son implantation globale. La structure bureaucratique et le manque de préparation du système socioculturel ont constitué les barrières les plus influentes sur: la diffusion de l’information; l’implication des acteurs du terrain dans le processus; et l’état de préparation du système éducationnel. Les principaux facilitateurs étaient : les valeurs promues à l’égard des droits humains et le mouvement politique pour renforcer le rôle professionnel de la sage-femme et réduire la mortalité maternelle. Quant au plan, il a été perçu comme étant bénéfique mais complexe et émanant d’une source externe. Les résultats mettent l’accent sur la nécessité de contourner les barrières identifiées dans les trois systèmes afin d’obtenir des contextes propices à la production des effets. Par ailleurs, les résultats ont soulevé aussi sept barrières qui risquent de compromettre l’atteinte des effets désirés. Elles concernent: le cadre légal, les représentations sociales et le support médiatique au niveau du système socioculturel; le réseautage et les mécanismes de communication, les caractéristiques liées au rôle, à l’environnement de pratique, et le niveau de préparation du système disciplinaire. Notre recherche confirme qu’un changement visant le système éducationnel isolément représente une vision réductrice pour le renforcement du rôle des sages-femmes. Une combinaison des conditions contextuelles favorables au niveau des dimensions des trois systèmes est requise pour atteindre le but de la stratégie gouvernementale, soit fournir des sages-femmes qualifiées selon les normes globales de la Confédération Internationale des sages-femmes, capables d’offrir des soins de qualité en santé de la reproduction qui permettront de contribuer à réduire la mortalité maternelle et néonatale.

Predicting violence in schizophrenia: a prospective study

Background: People with schizophrenia are more violent than the general population, but this increased risk is attributable to the actions of a small subgroup. Identifying those at risk has become an essential part of clinical practice. Aims: To estimate the risk factors for assault in patients with schizophrenia. Methods: Two hundred seventy-one patients with schizophrenia were interviewed using an extensive battery of instruments. Assault was measured from multiple data sources over the next 2 years and criminal records were obtained. Multiple sociodemographic and clinical variables measured at baseline were examined as possible predictors of assault during follow-up. Results: Sixty-nine (25%) patients committed assault during the 2-year followup. The model that best predicted assault included a history of recent assault (OR 2.33, 95% CI 1.17-4.61), a previous violent conviction (OR 2.02, 95% CI 1.04-3.87), having received special education (OR 2.76, 95% CI 1.22-6.26) and alcohol abuse (OR 3.55, 95% CI 1.24-10.2). Conclusions: Previously established risk factors including a history of violence and alcohol abuse are replicated in this study. Although low premorbid IQ did not predict violence, a need for special education did. (C) 2003 Published by Elsevier B.V.

Collaboration model for SSMEnetUK: a service science perspective

This paper aims to design a collaboration model for a Knowledge Community - SSMEnetUK. The research identifies SSMEnetUK as a socio-technical system and uses the core concepts of Service Science to explore the subject domain. The paper is positioned within the concept of Knowledge Management (KM) and utilising Web 2.0 tools for collaboration. A qualitative case study method was adopted and multiple data sources were used. In achieving that, the degree of co-relation between knowledge management activities and Web 2.0 tools for collaboration in the scenario are pitted against the concept of value propositions offered by both customer/user and service provider. The proposed model provides a better understanding of how Knowledge Management and Web 2.0 tools can enable effective collaboration within SSMEnetUK. This research is relevant to the wider service design and innovation community because it provides a basis for building a service-centric collaboration platform for the benefit of both customer/user and service provider.

Fatores determinantes na escolha do Brasil como exportador de serviços de tecnologia da informação

The export of information technology software services, also known as ¿offshore outsourcing¿, has raised debates in the media as well as in the academy. A lot has been written about the success of India, Ireland and Israel, the ¿3Is¿, but empirical data about Brazil is still hard to find. This dissertation proposes to identify success factors for Brazil to be chosen as a preferred location for offshore outsourcing based on a case study of an American multinational corporation, with branches in Brazil, that is systematically choosing Brazil as a preferred location for its offshore outsourcing operations. Concepts of economic globalization, internationalization of services and success factors for offshore outsourcing will be presented in the literature review and based on available literature focused on Brazil, a model of eight success factors is proposed. The empirical research was grounded on multiple data sources but the analysis was focused on a database of 219 deals that were conducted from September 2005 to May 2006, out of which Brazil was selected 57 times. The results confirm the proposed model of eight success factors. The final conclusions suggest that the process of identifying a country to perform the offshore activities is complex and that not all factors will be present at the same time, and more than that, in some cases intangible factors, such as relationship networks and emotional links with the country, have a higher weight in the decision. The results can be used in the future for in depth researches that differentiate Brazil from other countries in the offshore outsourcing market.

Residential self-selection and geographical scales in trip frequency.

Neighbourhood representation and scale used to measure the built environment have been treated in many ways. However, it is anything but clear what representation of neighbourhood is the most feasible in the existing literature. This paper presents an exhaustive analysis of built environment attributes through three spatial scales. For this purpose multiple data sources are integrated, and a set of 943 observations is analysed. This paper simultaneously analyses the influence of two methodological issues in the study of the relationship between built environment and travel behaviour: (1) detailed representation of neighbourhood by testing different spatial scales; (2) the influence of unobserved individual sensitivity to built environment attributes. The results show that different spatial scales of built environment attributes produce different results. Hence, it is important to produce local and regional transport measures, according to geographical scale. Additionally, the results show significant sensitivity to built environment attributes depending on place of residence. This effect, called residential sorting, acquires different magnitudes depending on the geographical scale used to measure the built environment attributes. Spatial scales risk to the stability of model results. Hence, transportation modellers and planners must take into account both effects of self-selection and spatial scales.

Beliefs, practices, and interactions of teachers in a Japanese high school English department

This study reveals the school culture and the teachers' professional development activities in a Japanese high school learning environment. Furthermore, it documents the relationships among the context, teachers' beliefs, practices, and interactions. Using multiple data sources including interviews, observations, and documents of teachers from an English department, this yearlong study revealed these English as a Foreign Language teachers lacked many teacher learning opportunities in their context. The study revealed that teacher collaboration only reinforced existing practices, eroding teachers' motivation to learn to teach in this specific context. The study provides evidence to teacher educators about inservice teachers and their learning environment and the significance of the relationships between the two entities. (C) 2004 Elsevier Ltd. All rights reserved.

A systems model for community injury prevention

The purpose of this study was to develop, explicate, and validate a comprehensive model in order to more effectively assess community injury prevention needs, plan and target efforts, identify potential interventions, and provide a framework for an outcome-based evaluation of the effectiveness of interventions. A systems model approach was developed to conceptualize the major components of inputs, efforts, outcomes and feedback within a community setting. Profiling of multiple data sources demonstrated a community feedback mechanism that increased awareness of priority issues and elicited support from traditional as well as non-traditional injury prevention partners. Injury countermeasures including education, enforcement, engineering, and economic incentives were presented for their potential synergistic effect impacting on knowledge, attitudes, or behaviors of a targeted population. Levels of outcome data were classified into ultimate, intermediate and immediate indicators to assist with determining the effectiveness of intervention efforts. A collaboration between business and health care was successful in achieving data access and use of an emergency department level of injury data for monitoring of the impact of community interventions. Evaluation of injury events and preventive efforts within the context of a dynamic community systems environment was applied to a study community with examples detailing actual profiling and trending of injuries. The resulting model of community injury prevention was validated using a community focus group, community injury prevention coordinators, and injury prevention national experts. ^

Evaluating the impacts of Bolsa Escola programs on student performance in Brazil

Since the 1980s, governments and organizations have promoted cash transfers in education as a tool for motivating elementary aged children to attend school. Oftentimes, the monthly payments supplemented the income a child would be making in the labor market. In Brazil, where these Bolsa or grant programs were pioneered, there has been much success in removing children from harsh labor conditions and increasing enrollment rates among the poorest families. However, the capacity of Bolsa Escola programs to meet other objectives, such as impacting educational outcomes and reducing incidences of poverty, continues to be examined. As these programs continue to be adopted globally, funding millions of children and families, evidence that demonstrates such success becomes ever more imperative. This study, therefore, examined evidence to determine whether Bolsa Escola programs have a significant impact on the academic performance of beneficiaries in Brazil. ^ Through the course of three data collection phases, multiple data sources were used to demonstrate the academic performance of fourth and eighth grade Brazilian students who were eligible to participate in either an NGO or the federal cash transfer program. MANOVAs were conducted separately for fourth and eighth grade data to determine if significant differences existed between measures of academic performance of Bolsa and non-Bolsa students. In every case and for both grade levels, significant effects were found for participation. ^ The limited qualitative data collected did not support drawing conclusions. Thematic analysis of the limited interview data pointed to possible dependency on Bolsa monthly stipends, and reallocation of responsibilities in the home in cases where children shifted from being breadwinners to students. ^

Bayesian Methods for Inferring Gene Regulatory Networks

Thesis (Ph.D.)--University of Washington, 2016-08

Bento Box Discovery

Poster presentation at the University of Maryland Libraries Research & Innovative Practice Forum on June 8, 2016. The poster proposes that the UMD Libraries should evaluate adoption of Bento Box Discovery for improved user search experience.

Security and Trust in Distributed Computation

We propose three research problems to explore the relations between trust and security in the setting of distributed computation. In the first problem, we study trust-based adversary detection in distributed consensus computation. The adversaries we consider behave arbitrarily disobeying the consensus protocol. We propose a trust-based consensus algorithm with local and global trust evaluations. The algorithm can be abstracted using a two-layer structure with the top layer running a trust-based consensus algorithm and the bottom layer as a subroutine executing a global trust update scheme. We utilize a set of pre-trusted nodes, headers, to propagate local trust opinions throughout the network. This two-layer framework is flexible in that it can be easily extensible to contain more complicated decision rules, and global trust schemes. The first problem assumes that normal nodes are homogeneous, i.e. it is guaranteed that a normal node always behaves as it is programmed. In the second and third problems however, we assume that nodes are heterogeneous, i.e, given a task, the probability that a node generates a correct answer varies from node to node. The adversaries considered in these two problems are workers from the open crowd who are either investing little efforts in the tasks assigned to them or intentionally give wrong answers to questions. In the second part of the thesis, we consider a typical crowdsourcing task that aggregates input from multiple workers as a problem in information fusion. To cope with the issue of noisy and sometimes malicious input from workers, trust is used to model workers' expertise. In a multi-domain knowledge learning task, however, using scalar-valued trust to model a worker's performance is not sufficient to reflect the worker's trustworthiness in each of the domains. To address this issue, we propose a probabilistic model to jointly infer multi-dimensional trust of workers, multi-domain properties of questions, and true labels of questions. Our model is very flexible and extensible to incorporate metadata associated with questions. To show that, we further propose two extended models, one of which handles input tasks with real-valued features and the other handles tasks with text features by incorporating topic models. Our models can effectively recover trust vectors of workers, which can be very useful in task assignment adaptive to workers' trust in the future. These results can be applied for fusion of information from multiple data sources like sensors, human input, machine learning results, or a hybrid of them. In the second subproblem, we address crowdsourcing with adversaries under logical constraints. We observe that questions are often not independent in real life applications. Instead, there are logical relations between them. Similarly, workers that provide answers are not independent of each other either. Answers given by workers with similar attributes tend to be correlated. Therefore, we propose a novel unified graphical model consisting of two layers. The top layer encodes domain knowledge which allows users to express logical relations using first-order logic rules and the bottom layer encodes a traditional crowdsourcing graphical model. Our model can be seen as a generalized probabilistic soft logic framework that encodes both logical relations and probabilistic dependencies. To solve the collective inference problem efficiently, we have devised a scalable joint inference algorithm based on the alternating direction method of multipliers. The third part of the thesis considers the problem of optimal assignment under budget constraints when workers are unreliable and sometimes malicious. In a real crowdsourcing market, each answer obtained from a worker incurs cost. The cost is associated with both the level of trustworthiness of workers and the difficulty of tasks. Typically, access to expert-level (more trustworthy) workers is more expensive than to average crowd and completion of a challenging task is more costly than a click-away question. In this problem, we address the problem of optimal assignment of heterogeneous tasks to workers of varying trust levels with budget constraints. Specifically, we design a trust-aware task allocation algorithm that takes as inputs the estimated trust of workers and pre-set budget, and outputs the optimal assignment of tasks to workers. We derive the bound of total error probability that relates to budget, trustworthiness of crowds, and costs of obtaining labels from crowds naturally. Higher budget, more trustworthy crowds, and less costly jobs result in a lower theoretical bound. Our allocation scheme does not depend on the specific design of the trust evaluation component. Therefore, it can be combined with generic trust evaluation algorithms.