Offline grammar-based recognition of handwritten sentences

This paper proposes a sequential coupling of a Hidden Markov Model (HMM) recognizer for offline handwritten English sentences with a probabilistic bottom-up chart parser using Stochastic Context-Free Grammars (SCFG) extracted from a text corpus. Based on extensive experiments, we conclude that syntax analysis helps to improve recognition rates significantly.

GIS based Watershed Classification in Lao PDR

Every inclined land surface has a potential for soil and water degradation, the seriousness depends on a multitude of parameters such as slope, soil type, geomorphology, rainfall, land use and natural vegetation cover. In Laos this intensified land use leads to reduced vegetation cover, to increased soil erosion, decreasing yield, and finally is likely to influence the hydrological regime. Against this background the Mekong River Commission (MRC) elaborated a spatial explicit Watershed Classification (WSC) for the Lower Mekong Basin. Based on topographic factors derived from a high-resolution Digital Terrain Model, five watershed classes are calculated, giving indication about the sensitivity to resource degradation by soil erosion. The WSC allows spatial priority setting for watershed management and generally supports informed decision making on reconnaissance level. In the conclusions focus is laid on general considerations when GIS techniques are used for spatial decision support in a development context.

Understanding Research Fatigue in the Context of Community-University Relations

Community research fatigue has been understudied within the context of community-university relationships and knowledge production. Community-based research (CBR), often occurring within a limited geography and population, increases the possibility that community members feel exhausted or over-whelmed by university research —particularly when they do not see tangible results from research activities. Prompted by informal stories of research fatigue from community members, a small graduate student team sought to understand the extent to which community members experienced research fatigue, and what factors contributed to or relieved feelings of research fatigue. In order to explore these dimensions of research fatigue, semi-structured, face-to-face interviews were conducted with 21 participants, including community members (n = 9), staff and faculty (n = 10), and students (n = 2). The objective of the research was to identify university practices that contribute to research fatigue and how to address the issue at the university level. Qualitative data analysis revealed several important actionable findings: the structure and conduct of community-based research, structured reciprocity and impact, and the role of trust in research. This study’s findings are used to assess the quality of Clark University’s research relationship with its adjacent community. Recommendations are offered; such as to improve partnerships, the impact of CBR, and to develop clear principles of practice.

[MRI-based diagnosis of giant cell arteritis]

The giant cell arteritis and its symptoms are usually non-specific and accompanied with symptoms of polymyalgia rheumatica. As complications of the giant cell arteritis ischemia, infarction or rupture of the damaged vessel can occur. We report on a 56-year-old female patient, who suffered for one year about weight loss, tiredness and intolerance as well as symptoms of polymyalgia rheumatica. Gastroscopy and colonoscopy showed normal findings. In the context of the malignancy search we made a computer tomography and magnet resonance tomography. The data showed an enlargement and an enhancement of the aorta, which led us to the suspicion of a giant cell arteritis. We started immediately with a medical treatment. The biopsy of the arteries temporales supported histological the diagnosis.

A Functional-Based Distribution Diagnostic for a Linear Model with Correlated Outcomes: Technical Report

Despite the widespread popularity of linear models for correlated outcomes (e.g. linear mixed modesl and time series models), distribution diagnostic methodology remains relatively underdeveloped in this context. In this paper we present an easy-to-implement approach that lends itself to graphical displays of model fit. Our approach involves multiplying the estimated marginal residual vector by the Cholesky decomposition of the inverse of the estimated marginal variance matrix. Linear functions or the resulting "rotated" residuals are used to construct an empirical cumulative distribution function (ECDF), whose stochastic limit is characterized. We describe a resampling technique that serves as a computationally efficient parametric bootstrap for generating representatives of the stochastic limit of the ECDF. Through functionals, such representatives are used to construct global tests for the hypothesis of normal margional errors. In addition, we demonstrate that the ECDF of the predicted random effects, as described by Lange and Ryan (1989), can be formulated as a special case of our approach. Thus, our method supports both omnibus and directed tests. Our method works well in a variety of circumstances, including models having independent units of sampling (clustered data) and models for which all observations are correlated (e.g., a single time series).

Bayes Factors Based on Test Statistics

Traditionally, the use of Bayes factors has required the specification of proper prior distributions on model parameters implicit to both null and alternative hypotheses. In this paper, I describe an approach to defining Bayes factors based on modeling test statistics. Because the distributions of test statistics do not depend on unknown model parameters, this approach eliminates the subjectivity normally associated with the definition of Bayes factors. For standard test statistics, including the _2, F, t and z statistics, the values of Bayes factors that result from this approach can be simply expressed in closed form.

A unifying approach for haplotype analysis of quantitative traits in family-based association studies: Testing and estimating gene-environment interactions with complex exposure variables

We propose robust and e±cient tests and estimators for gene-environment/gene-drug interactions in family-based association studies. The methodology is designed for studies in which haplotypes, quantitative pheno- types and complex exposure/treatment variables are analyzed. Using causal inference methodology, we derive family-based association tests and estimators for the genetic main effects and the interactions. The tests and estimators are robust against population admixture and strati¯cation without requiring adjustment for confounding variables. We illustrate the practical relevance of our approach by an application to a COPD study. The data analysis suggests a gene-environment interaction between a SNP in the Serpine gene and smok- ing status/pack years of smoking that reduces the FEV1 volume by about 0.02 liter per pack year of smoking. Simulation studies show that the pro- posed methodology is su±ciently powered for realistic sample sizes and that it provides valid tests and effect size estimators in the presence of admixture and stratification.

Tracking the subprocesses of decision-based action in the human frontal lobes

Situationally adaptive behavior relies on the identification of relevant target stimuli, the evaluation of these with respect to the current context and the selection of an appropriate action. We used functional magnetic resonance imaging (fMRI) to disentangle the neural networks underlying these processes within a single task. Our results show that activation of mid-ventrolateral prefrontal cortex (PFC) reflects the perceived presence of a target stimulus regardless of context, whereas context-appropriate evaluation is subserved by mid-dorsolateral PFC. Enhancing demands on response selection by means of response conflict activated a network of regions, all of which are directly connected to motor areas. On the midline, rostral anterior paracingulate cortex was found to link target detection and response selection by monitoring for the presence of behaviorally significant conditions. In summary, we provide new evidence for process-specific functional dissociations in the frontal lobes. In target-centered processing, target detection in the VLPFC is separable from contextual evaluation in the DLPFC. Response-centered processing in motor-associated regions occurs partly in parallel to these processes, which may enhance behavioral efficiency, but it may also lead to reaction time increases when an irrelevant response tendency is elicited.

Model Evaluation Based on the Distribution of Estimated Absolute Prediction Error

The construction of a reliable, practically useful prediction rule for future response is heavily dependent on the "adequacy" of the fitted regression model. In this article, we consider the absolute prediction error, the expected value of the absolute difference between the future and predicted responses, as the model evaluation criterion. This prediction error is easier to interpret than the average squared error and is equivalent to the mis-classification error for the binary outcome. We show that the distributions of the apparent error and its cross-validation counterparts are approximately normal even under a misspecified fitted model. When the prediction rule is "unsmooth", the variance of the above normal distribution can be estimated well via a perturbation-resampling method. We also show how to approximate the distribution of the difference of the estimated prediction errors from two competing models. With two real examples, we demonstrate that the resulting interval estimates for prediction errors provide much more information about model adequacy than the point estimates alone.

Inference on Survival Data with Covariate Measurement Error - An Imputation-based Approach

We propose a new method for fitting proportional hazards models with error-prone covariates. Regression coefficients are estimated by solving an estimating equation that is the average of the partial likelihood scores based on imputed true covariates. For the purpose of imputation, a linear spline model is assumed on the baseline hazard. We discuss consistency and asymptotic normality of the resulting estimators, and propose a stochastic approximation scheme to obtain the estimates. The algorithm is easy to implement, and reduces to the ordinary Cox partial likelihood approach when the measurement error has a degenerative distribution. Simulations indicate high efficiency and robustness. We consider the special case where error-prone replicates are available on the unobserved true covariates. As expected, increasing the number of replicate for the unobserved covariates increases efficiency and reduces bias. We illustrate the practical utility of the proposed method with an Eastern Cooperative Oncology Group clinical trial where a genetic marker, c-myc expression level, is subject to measurement error.