Minimum description length, regularisation and multi-modal data

Conventional feed forward Neural Networks have used the sum-of-squares cost function for training. A new cost function is presented here with a description length interpretation based on Rissanen's Minimum Description Length principle. It is a heuristic that has a rough interpretation as the number of data points fit by the model. Not concerned with finding optimal descriptions, the cost function prefers to form minimum descriptions in a naive way for computational convenience. The cost function is called the Naive Description Length cost function. Finding minimum description models will be shown to be closely related to the identification of clusters in the data. As a consequence the minimum of this cost function approximates the most probable mode of the data rather than the sum-of-squares cost function that approximates the mean. The new cost function is shown to provide information about the structure of the data. This is done by inspecting the dependence of the error to the amount of regularisation. This structure provides a method of selecting regularisation parameters as an alternative or supplement to Bayesian methods. The new cost function is tested on a number of multi-valued problems such as a simple inverse kinematics problem. It is also tested on a number of classification and regression problems. The mode-seeking property of this cost function is shown to improve prediction in time series problems. Description length principles are used in a similar fashion to derive a regulariser to control network complexity.

Low-income first time mothers: Effects on advanced practice nurse (APN) follow up telephone calls on maternal and infant health and health care charges

The United States has over 4 million births annually. Currently healthy women with non-complicated deliveries receive little to no routine postpartum support when discharged from the hospital. This is especially problematic if mothers are first time mothers, poor, have language barriers and little to no social support after giving birth. The purpose of this randomized clinical trial was to compare maternal and infant health outcomes, and health care charges between 2 groups of mothers and newborns. A control ( n = 69) group received routine posthospital discharge care. An intervention group (n = 70) received routine posthospital discharge care plus follow up telephone calls by advanced practice nurses (APNs) on days 3,7,14,21,28 and week 8. Both groups were followed for the first 8 weeks posthospital discharge following delivery to examine maternal health outcomes (perceived maternal stress, social support and perceived maternal physical health), infant health outcomes (routine medical follow up visits immunizations, weight gain), morbidity (urgent care visits, emergency room visits, rehospitalizations), health care charges (urgent care visits, emergency room visits, rehospitalizations) in both groups and charges for APN follow up in the intervention group only. Data were analyzed using descriptive statistics and two-sample t-tests. Study findings indicated that intervention group had significantly lower perceived maternal stress, significantly higher rating of perceived maternal health and higher levels of social support and by the end of the 2nd month posthospital discharge compared to control group mothers. Infants in the intervention group had: increased number of immunizations; fewer emergency room visits; and 1 infant rehospitalization compared to 3 infant rehospitalizations in the control group. The intervention groups' health care charges were significantly lower compared to the control group $14,333/$497 vs. $70,834/$1,068. These study results indicate that an intervention of APN follow up telephone calls in this sample of first time low-income culturally diverse mothers was an effective, safe, low cost, easy to apply intervention which improved mothers' and infants' health outcomes and reduced healthcare charges.

The Real Failure Rate of Restaurants

A common assumption in the restaurant industry is that restaurants fail at an exceedingly high rate. However, statistical research to support this assumption is limited. The authors present a study of 10 years in the life of three markets and offer new data for managers to consider.

Resistance is futile: the borg, the hive, and corporate hegemony

The Borg, a collective of humanoid cyborgs linked together in a hive-mind and modeled on the earthly superorganisms of ant colonies and beehives, has been the most feared alien race in the Star Trek universe. The formidable success of the Borg in assimilating their foes corresponds to the astounding success of superorganisms in our own biosphere. Yet the Borg also serves as a metaphor for another collective of biological entities known as the corporation. In the Anthropocene epoch, corporations have become the most powerful force on the planet; their influence on the social world and the environment exceeds any government and may determine the continued sustainability of human life. Corporations have been described as people and as machines, but neither metaphor accurately describes their essence or contributes to an understanding that might resist their power. This paper reframes our understanding of the corporation by examining the metaphors that are used to describe it, and by suggesting an entirely new metaphor viewing the Borg and the corporation through the lens of sociobiology. I will argue that the corporation is a new form of superorganism that has become the dominant species on the planet and that the immense, intractable power of a globalized, corporate hive-mind has become the principal obstacle to addressing the planetary emergency of climate change. Reframing our metaphoric understanding of corporations as biological entities in the planetary biosphere may enable us to imagine ways to resist their increasing dominance and create a sustainable future. 

A Figured Worlds Perspective on Middle School Learners' Climate Literacy Development

The purpose of this study was to investigate the nature of the relationship between middle school science learners’ conditions and their developing understandings of climate change. I applied the anthropological theoretical perspective of figured worlds (Holland, Lachicotte, Skinner, & Cain, 1998) to examine learners’ views of themselves and their capacities to act in relation to climate change. My overarching research question was: How are middle school science learners’ figured worlds of climate change related to the conditions in which they are embedded? I used a descriptive single-case study design to examine the climate change ideas of eight purposefully selected 6th grade science learners. Data sources included: classroom observations, curriculum documents, interviews, focus groups, and written assessments and artifacts, including learners’ self- generated drawings. I identified six analytic lenses with which to explore the data. Insights from the application of these analytic lenses provided information about the elements of participants’ climate change stories, which I reported through the use of a storytelling heuristic. I then synthesized elements of participants’ collective climate change story, which provided an “entrance” (Kitchell, Hannan, & Kempton, 2000, p. 96) into their figured world of climate change. Aspects of learners’ conditions—such as their worlds of school, technology and media use, and family—appeared to shape their figured world of climate change. Within their figured world of climate change, learners saw themselves—individually and as members of groups—as inhabiting a variety of climate change identities, some of which were in conflict with each other. I posited that learners’ enactment of these identities – or the ways in which they expressed their climate change agency – had the potential to reshape or reinforce their conditions. Thus, learners’ figured worlds of climate change might be considered “spaces of authoring” (Holland et al., 1998, p. 45) with potential for inciting social and environmental change. The nature of such change would hinge on the extent to which these nascent climate change identities become salient for these early adolescent learners through their continued climate change learning experiences. Implications for policy, curriculum and instruction, and science education research related to climate change education are presented.

O comprimento descritivo mínimo na amostragem por transectos lineares

Apresenta·se um breve resumo histórico da evolução da amostragem por transectos lineares e desenvolve·se a sua teoria. Descrevemos a teoria de amostragem por transectos lineares, proposta por Buckland (1992), sendo apresentados os pontos mais relevantes, no que diz respeito à modelação da função de detecção. Apresentamos uma descrição do princípio CDM (Rissanen, 1978) e a sua aplicação à estimação de uma função densidade por um histograma (Kontkanen e Myllymãki, 2006), procedendo à aplicação de um exemplo prático, recorrendo a uma mistura de densidades. Procedemos à sua aplicação ao cálculo do estimador da probabilidade de detecção, no caso dos transectos lineares e desta forma estimar a densidade populacional de animais. Analisamos dois casos práticos, clássicos na amostragem por distâncias, comparando os resultados obtidos. De forma a avaliar a metodologia, simulámos vários conjuntos de observações, tendo como base o exemplo das estacas, recorrendo às funções de detecção semi-normal, taxa de risco, exponencial e uniforme com um cosseno. Os resultados foram obtidos com o programa DISTANCE (Thomas et al., in press) e um algoritmo escrito em linguagem C, cedido pelo Professor Doutor Petri Kontkanen (Departamento de Ciências da Computação, Universidade de Helsínquia). Foram desenvolvidos programas de forma a calcular intervalos de confiança recorrendo à técnica bootstrap (Efron, 1978). São discutidos os resultados finais e apresentadas sugestões de desenvolvimentos futuros. ABSTRACT; We present a brief historical note on the evolution of line transect sampling and its theoretical developments. We describe line transect sampling theory as proposed by Buckland (1992), and present the most relevant issues about modeling the detection function. We present a description of the CDM principle (Rissanen, 1978) and its application to histogram density estimation (Kontkanen and Myllymãki, 2006), with a practical example, using a mixture of densities. We proceed with the application and estimate probability of detection and animal population density in the context of line transect sampling. Two classical examples from the literature are analyzed and compared. ln order to evaluate the proposed methodology, we carry out a simulation study based on a wooden stakes example, and using as detection functions half normal, hazard rate, exponential and uniform with a cosine term. The results were obtained using program DISTANCE (Thomas et al., in press), and an algorithm written in C language, kindly offered by Professor Petri Kontkanen (Department of Computer Science, University of Helsinki). We develop some programs in order to estimate confidence intervals using the bootstrap technique (Efron, 1978). Finally, the results are presented and discussed with suggestions for future developments.

A comparison of collection techniques for gene expression analysis of human oral taste tissue

Variability in human taste perception is associated with both genetic and environmental factors. The influence of taste receptor expression on this variability is unknown, in part, due to the difficulty in obtaining human oral tissue that enables quantitative expression measures of taste genes. In a comparison of six current techniques (Oragene RNeasy Kit, Isohelix swab, Livibrush cytobrush, tongue saliva, cheek saliva collection, and fungiform papillae biopsy), we identify the fungiform papillae biopsy is the optimal sampling technique to analyse human taste gene expression. The fungiform papillae biopsy resulted in the highest RNA integrity, enabling amplification of all the assessed taste receptor genes (TAS1R1, TAS1R2, TAS1R3, SCNN1A and CD36) and taste tissue marker genes (NCAM1, GNAT3 and PLCβ2). Furthermore, quantitative expression was observed in a subset of taste genes assessed from the saliva collection techniques (cheek saliva, tongue saliva and Oragene RNA kit). These saliva collection techniques may be useful as a non-invasive alternative sampling technique to the fungiform papillae biopsy. Identification of the fungiform papillae biopsy as the optimal collection method will facilitate further research into understanding the effect of gene expression on variability in human taste perception.

N-acetylcysteine modulates glutamatergic dysfunction and depressive behavior in Huntington's disease

Glutamatergic dysfunction has been implicated in the pathogenesis of depressive disorders and Huntington's disease (HD), in which depression is the most common psychiatric symptom. Synaptic glutamate homeostasis is regulated by cystine-dependent glutamate transporters, including GLT-1 and system xc (-) In HD, the enzyme regulating cysteine (and subsequently cystine) production, cystathionine-γ-lygase, has recently been shown to be lowered. The aim of the present study was to establish whether cysteine supplementation, using N-acetylcysteine (NAC) could ameliorate glutamate pathology through the cystine-dependent transporters, system xc (-) and GLT-1. We demonstrate that the R6/1 transgenic mouse model of HD has lower basal levels of cystine, and showed depressive-like behaviors in the forced-swim test. Administration of NAC reversed these behaviors. This effect was blocked by co-administration of the system xc (-) and GLT-1 inhibitors CPG and DHK, showing that glutamate transporter activity was required for the antidepressant effects of NAC. NAC was also able to specifically increase glutamate in HD mice, in a glutamate transporter-dependent manner. These in vivo changes reflect changes in glutamate transporter protein in HD mice and human HD post-mortem tissue. Furthermore, NAC was able to rescue changes in key glutamate receptor proteins related to excitotoxicity in HD, including NMDAR2B. Thus, we have shown that baseline reductions in cysteine underlie glutamatergic dysfunction and depressive-like behavior in HD and these changes can be rescued by treatment with NAC. These findings have implications for the development of new therapeutic approaches for depressive disorders.

Impact of early valve surgery on outcome of Staphylococcus aureus prosthetic valve infective endocarditis: analysis in the International Collaboration of Endocarditis-Prospective Cohort Study

BACKGROUND: The impact of early valve surgery (EVS) on the outcome of Staphylococcus aureus (SA) prosthetic valve infective endocarditis (PVIE) is unresolved. The objective of this study was to evaluate the association between EVS, performed within the first 60 days of hospitalization, and outcome of SA PVIE within the International Collaboration on Endocarditis-Prospective Cohort Study. METHODS: Participants were enrolled between June 2000 and December 2006. Cox proportional hazards modeling that included surgery as a time-dependent covariate and propensity adjustment for likelihood to receive cardiac surgery was used to evaluate the impact of EVS and 1-year all-cause mortality on patients with definite left-sided S. aureus PVIE and no history of injection drug use. RESULTS: EVS was performed in 74 of the 168 (44.3%) patients. One-year mortality was significantly higher among patients with S. aureus PVIE than in patients with non-S. aureus PVIE (48.2% vs 32.9%; P = .003). Staphylococcus aureus PVIE patients who underwent EVS had a significantly lower 1-year mortality rate (33.8% vs 59.1%; P = .001). In multivariate, propensity-adjusted models, EVS was not associated with 1-year mortality (risk ratio, 0.67 [95% confidence interval, .39-1.15]; P = .15). CONCLUSIONS: In this prospective, multinational cohort of patients with S. aureus PVIE, EVS was not associated with reduced 1-year mortality. The decision to pursue EVS should be individualized for each patient, based upon infection-specific characteristics rather than solely upon the microbiology of the infection causing PVIE.

Efectos de una intervención educativa en la promoción de la actividad física, otros comportamientos saludables y los conocimientos para la prevención del cáncer de pulmón en jóvenes del colegio La Merced en Bogotá D.C., Colombia

Introducción: El cáncer de pulmón es el tipo de cáncer más mortal a nivel mundial, al cual se atribuyen una de cada cinco muertes anualmente. El objetivo de este estudio fue determinar los efectos de una intervención educativa en la promoción de la actividad física, otros comportamientos saludables y los conocimientos para la prevención del cáncer de pulmón en jóvenes estudiantes de una institución educativa pública en Bogotá, Colombia. Métodos: Estudio experimental no controlado en 243 estudiantes de sexo femenino (Edad 14±1,5 años). La intervención educativa se desarrolló en tres momentos: una sesión educativa con una duración de 60 minutos acorde a la Guía para la Comunicación Educativa en el Marco del Control del Cáncer, en Colombia. Segundo, se enviaron tres correos electrónicos con información acerca del cáncer pulmonar; finalmente se desarrollaron actividades grupales. Para la toma de datos se utilizaron el cuestionario Cáncer Awareness Measure (CAM) y el Sistema de Vigilancia de Factores de Riesgo del Comportamiento (BRFSS). La evaluación se realizó en un período de seguimiento a 1, 3 y 6 meses post-intervención. Resultados: La intervención educativa incrementó significativamente los conocimientos de las jóvenes sobre los signos de alarma del cáncer pulmonar y los principales factores de riesgo modificables, tales como el consumo de cigarrillo, la exposición al humo del mismo y el sedentarismo, al sexto mes post-intervención. Las mejoras en el cambio comportamental no lograron significancia estadística. Conclusiones: Una intervención educativa mejora los conocimientos acerca de la detección temprana y la prevención del cáncer de pulmón, así como los comportamientos saludables en jóvenes estudiantes en Bogotá, Colombia. Se requiere de estudios controlados aleatorizados.