Squared Extrapolation Methods (SQUAREM): A New Class of Simple and Efficient Numerical Schemes for Accelerating the Convergence of the EM Algorithm

We derive a new class of iterative schemes for accelerating the convergence of the EM algorithm, by exploiting the connection between fixed point iterations and extrapolation methods. First, we present a general formulation of one-step iterative schemes, which are obtained by cycling with the extrapolation methods. We, then square the one-step schemes to obtain the new class of methods, which we call SQUAREM. Squaring a one-step iterative scheme is simply applying it twice within each cycle of the extrapolation method. Here we focus on the first order or rank-one extrapolation methods for two reasons, (1) simplicity, and (2) computational efficiency. In particular, we study two first order extrapolation methods, the reduced rank extrapolation (RRE1) and minimal polynomial extrapolation (MPE1). The convergence of the new schemes, both one-step and squared, is non-monotonic with respect to the residual norm. The first order one-step and SQUAREM schemes are linearly convergent, like the EM algorithm but they have a faster rate of convergence. We demonstrate, through five different examples, the effectiveness of the first order SQUAREM schemes, SqRRE1 and SqMPE1, in accelerating the EM algorithm. The SQUAREM schemes are also shown to be vastly superior to their one-step counterparts, RRE1 and MPE1, in terms of computational efficiency. The proposed extrapolation schemes can fail due to the numerical problems of stagnation and near breakdown. We have developed a new hybrid iterative scheme that combines the RRE1 and MPE1 schemes in such a manner that it overcomes both stagnation and near breakdown. The squared first order hybrid scheme, SqHyb1, emerges as the iterative scheme of choice based on our numerical experiments. It combines the fast convergence of the SqMPE1, while avoiding near breakdowns, with the stability of SqRRE1, while avoiding stagnations. The SQUAREM methods can be incorporated very easily into an existing EM algorithm. They only require the basic EM step for their implementation and do not require any other auxiliary quantities such as the complete data log likelihood, and its gradient or hessian. They are an attractive option in problems with a very large number of parameters, and in problems where the statistical model is complex, the EM algorithm is slow and each EM step is computationally demanding.

Influence of Information and Instructions on Human Behavior in Tunnel Accidents: A Virtual Reality Study

Human behavior is a major factor modulating the consequences of road tunnel accidents. We investigated the effect of information and instruction on drivers' behavior as well as the usability of virtual environments to simulate such emergency situations. Tunnel safety knowledge of the general population was assessed using an online questionnaire, and tunnel safety behavior was investigated in a virtual reality experiment. Forty-four participants completed three drives through a virtual road tunnel and were confronted with a traffic jam, no event, and an accident blocking the road. Participants were randomly assigned to a control group (no intervention), an informed group who read a brochure containing safety information prior to the tunnel drives, or an informed and instructed group who read the same brochure and received additional instructions during the emergency situation. Informed participants showed better and quicker safety behavior than the control group. Self-reports of anxiety were assessed three times during each drive. Anxiety was elevated during and after the emergency situation. The findings demonstrate problematic safety behavior in the control group and that knowledge of safety information fosters adequate behavior in tunnel emergencies. Enhanced anxiety ratings during the emergency situation indicate external validity of the virtual environment.

Breast cancer on the world wide web: cross sectional survey of quality of information and popularity of websites.

OBJECTIVES: To determine the characteristics of popular breast cancer related websites and whether more popular sites are of higher quality. DESIGN: The search engine Google was used to generate a list of websites about breast cancer. Google ranks search results by measures of link popularity---the number of links to a site from other sites. The top 200 sites returned in response to the query "breast cancer" were divided into "more popular" and "less popular" subgroups by three different measures of link popularity: Google rank and number of links reported independently by Google and by AltaVista (another search engine). MAIN OUTCOME MEASURES: Type and quality of content. RESULTS: More popular sites according to Google rank were more likely than less popular ones to contain information on ongoing clinical trials (27% v 12%, P=0.01 ), results of trials (12% v 3%, P=0.02), and opportunities for psychosocial adjustment (48% v 23%, P<0.01). These characteristics were also associated with higher number of links as reported by Google and AltaVista. More popular sites by number of linking sites were also more likely to provide updates on other breast cancer research, information on legislation and advocacy, and a message board service. Measures of quality such as display of authorship, attribution or references, currency of information, and disclosure did not differ between groups. CONCLUSIONS: Popularity of websites is associated with type rather than quality of content. Sites that include content correlated with popularity may best meet the public's desire for information about breast cancer.

The dicode workbench: A flexible framework for the integration of information and web services

Aiming to address requirements concerning integration of services in the context of ?big data?, this paper presents an innovative approach that (i) ensures a flexible, adaptable and scalable information and computation infrastructure, and (ii) exploits the competences of stakeholders and information workers to meaningfully confront information management issues such as information characterization, classification and interpretation, thus incorporating the underlying collective intelligence. Our approach pays much attention to the issues of usability and ease-of-use, not requiring any particular programming expertise from the end users. We report on a series of technical issues concerning the desired flexibility of the proposed integration framework and we provide related recommendations to developers of such solutions. Evaluation results are also discussed.

The Future of the Technologies of Shell and spatial Structures. Bridges

The objective of this lecture is try to predict the future of this important type of spatial structures. In this way the activities of the different IASS Technical Working Groups can be stimulated and coordinated in order to play a more relevant role in this future. To grasp a possible evolution of bridges it is convenient a reflection on the bridge history and on their present situation, particularly in relation to the different existing achievements.

Desarrollo de una plataforma de biología bajo demanda

La Biología bajo Demanda es un concepto novedoso, que está siendo abordado en la actualidad desde distintos enfoques, que serán expuestos en este documento. Dado este carácter innovador, se trata de un ámbito donde la investigación está muy presente en estos momentos. Las Tecnologías de la Información y Comunicación (TICs) llevan un tiempo aportando soluciones muy efectivas para algunos de los problemas a los que se enfrente actualmente la biología sintética. Una de estas soluciones son las plataformas de Cloud Computing, que aportan un entorno de trabajo escalable, flexible y seguro. Por ello, se ha empleado este tipo de tecnología en este trabajo fin de grado en el área de la biología sintética mediante el concepto de biología bajo demanda. Para desarrollar la plataforma de biología bajo demanda ha sido necesario analizar el estado de esta temática actualmente y sus avances. Además, ha sido estimable el estudio de las opiniones de los miembros del grupo de investigación. Todo ello ha permitido llevar a cabo una captura de requisitos adecuada para el ámbito de este proyecto. Se ha decidido que los servidores de aplicaciones web son la respuesta más adecuada a la hora de implementar las soluciones obtenidas para el desarrollo de la plataforma de biología bajo demanda. En concreto, por sus características, se ha decidido emplear JavaEE de Oracle. El modelo implementado emplea soluciones conocidas y fiables basadas en patrones de diseño software. Así, conseguimos cumplir con uno de los principales objetivos de este proyecto, que es lograr un sistema flexible y escalable. Por otro lado, debido a la incertidumbre que conlleva un área tan innovadora, se ha decidido optar por una metodología ágil. Esto supone un plan de trabajo centrado en reuniones semanales conjuntas con el director y los compañeros del grupo de trabajo, empleando prototipado rápido y programación extrema. Finalmente, se ha conseguido desarrollar una plataforma de biología bajo demanda que puede ser la base para el trabajo de los biólogos del ámbito de la biología sintética en un futuro próximo.---ABSTRACT---Biology on demand is a new concept, which is currently being addressed from different approaches, which will be presented in this document. Given this innovative character, it is an area where research is a main factor right now. Technologies of Information and Communication Technologies (ICTs) have provided very effective solutions to some of the problems that synthetic biology is currently facing. One of these solutions is cloud computing platforms, which provide an environment for scalable, flexible and secure work. Therefore, we have used this technology in this final project in the area of synthetic biology through the concept of biology on demand. To develop a biology-on-demand platform it has been necessary to analyze the state of art. The opinions of members of the research group have also been very influential. All this has allowed us to conduct a proper capture requirements for the scope of this project here developed. It was decided that web application servers are the best answer when it comes to implementing the solutions obtained for the development of biology-on-demand platform. In particular, by its main features, it was decided to use Oracle’s JavaEE. The implemented model uses known and reliable solutions based on software design patterns. So, we get to meet one of the main objectives of this project, which is to achieve a flexible and scalable system. On the other hand, due to the uncertainty involved in such an innovative area, it was appropriate to opt for an agile methodology. The work plan was focused on weekly meetings with the director and coworkers, using additive technology and extreme programming. Finally, this project has been successful in developing a biology-on-demand platform that can be the basis for the work of biologists in the field of synthetic biology in the near future.

Notes on Philippine mosquitoes, XIII. Four new species of Zeugnomyia and Topomyia [by] Francisco E. Baisas and Pablo Feliciano.

v.33:no.3(1953)

New uses of celluloid and similar material in taxidermy / by Leon L. Walters --; Wilfred H. Osgood -- editor.

no.2(1925)