Hierarchical conditional random fields for parts based models matching

A parts based model is a parametrization of an object class using a collection of landmarks following the object structure. The matching of parts based models is one of the problems where pairwise Conditional Random Fields have been successfully applied. The main reason of their effectiveness is tractable inference and learning due to the simplicity of involved graphs, usually trees. However, these models do not consider possible patterns of statistics among sets of landmarks, and thus they sufffer from using too myopic information. To overcome this limitation, we propoese a novel structure based on a hierarchical Conditional Random Fields, which we explain in the first part of this memory. We build a hierarchy of combinations of landmarks, where matching is performed taking into account the whole hierarchy. To preserve tractable inference we effectively sample the label set. We test our method on facial feature selection and human pose estimation on two challenging datasets: Buffy and MultiPIE. In the second part of this memory, we present a novel approach to multiple kernel combination that relies on stacked classification. This method can be used to evaluate the landmarks of the parts-based model approach. Our method is based on combining responses of a set of independent classifiers for each individual kernel. Unlike earlier approaches that linearly combine kernel responses, our approach uses them as inputs to another set of classifiers. We will show that we outperform state-of-the-art methods on most of the standard benchmark datasets.

An empirical hierarchical Bayesian unification of occupational exposure assessment methods.

In occupational exposure assessment of airborne contaminants, exposure levels can either be estimated through repeated measurements of the pollutant concentration in air, expert judgment or through exposure models that use information on the conditions of exposure as input. In this report, we propose an empirical hierarchical Bayesian model to unify these approaches. Prior to any measurement, the hygienist conducts an assessment to generate prior distributions of exposure determinants. Monte-Carlo samples from these distributions feed two level-2 models: a physical, two-compartment model, and a non-parametric, neural network model trained with existing exposure data. The outputs of these two models are weighted according to the expert's assessment of their relevance to yield predictive distributions of the long-term geometric mean and geometric standard deviation of the worker's exposure profile (level-1 model). Bayesian inferences are then drawn iteratively from subsequent measurements of worker exposure. Any traditional decision strategy based on a comparison with occupational exposure limits (e.g. mean exposure, exceedance strategies) can then be applied. Data on 82 workers exposed to 18 contaminants in 14 companies were used to validate the model with cross-validation techniques. A user-friendly program running the model is available upon request.

Appropriateness of respiratory care: development of evidence-based guidelines in an acute care hospital

Background: Respiratory care is universally recognised as useful, but its indications and practice vary markedly. In order to improve appropriateness of respiratory care in our hospital, we developed evidence-based local guidelines in a collaborative effort involving physiotherapists, physicians, and health services researchers. Methods: Recommendations were developed using the standardised RAND appropriateness method. A literature search was performed for the period between 1995 and 2008 based on terms associated with guidelines and with respiratory care. Publications were assessed according to the Oxford classification of quality of evidence. A working group prepared proposals for recommendations which were then independently rated by a multidisciplinary expert panel. All recommendations were then discussed in common and indications for procedures were rated confidentially a second time by the experts. Each indication for respiratory care was classified as appropriate, uncertain, or inappropriate, based on the panel median rating and the degree of intra-panel agreement. Results: Recommendations were formulated for the following procedures: non-invasive ventilation, continuous positive airway pressure, intermittent positive pressure breathing, intrapulmonary percussive ventilation, mechanical insufflation-exsufflation, incentive spirometry, positive expiratory pressure, nasotracheal suctioning, noninstrumental airway clearance techniques. Each recommendation referred to a particular medical condition, and was assigned to a hierarchical category based on the quality of evidence from literature supporting the recommendation and on the consensus of experts. Conclusion: Despite a marked heterogeneity of scientific evidence, the method used allowed us to develop commonly agreed local guidelines for respiratory care. In addition, this work fostered a closer relationship between physiotherapists and physicians in our institution.

Evaluation of email alerts in practice: Part 2. Validation of the information assessment method.

RATIONALE AND OBJECTIVE:. The information assessment method (IAM) permits health professionals to systematically document the relevance, cognitive impact, use and health outcomes of information objects delivered by or retrieved from electronic knowledge resources. The companion review paper (Part 1) critically examined the literature, and proposed a 'Push-Pull-Acquisition-Cognition-Application' evaluation framework, which is operationalized by IAM. The purpose of the present paper (Part 2) is to examine the content validity of the IAM cognitive checklist when linked to email alerts. METHODS: A qualitative component of a mixed methods study was conducted with 46 doctors reading and rating research-based synopses sent on email. The unit of analysis was a doctor's explanation of a rating of one item regarding one synopsis. Interviews with participants provided 253 units that were analysed to assess concordance with item definitions. RESULTS AND CONCLUSION: The content relevance of seven items was supported. For three items, revisions were needed. Interviews suggested one new item. This study has yielded a 2008 version of IAM.

Method to assess and optimise dependability of complex macro-systems: application to a railway signalling system

Achieving a high degree of dependability in complex macro-systems is challenging. Because of the large number of components and numerous independent teams involved, an overview of the global system performance is usually lacking to support both design and operation adequately. A functional failure mode, effects and criticality analysis (FMECA) approach is proposed to address the dependability optimisation of large and complex systems. The basic inductive model FMECA has been enriched to include considerations such as operational procedures, alarm systems. environmental and human factors, as well as operation in degraded mode. Its implementation on a commercial software tool allows an active linking between the functional layers of the system and facilitates data processing and retrieval, which enables to contribute actively to the system optimisation. The proposed methodology has been applied to optimise dependability in a railway signalling system. Signalling systems are typical example of large complex systems made of multiple hierarchical layers. The proposed approach appears appropriate to assess the global risk- and availability-level of the system as well as to identify its vulnerabilities. This enriched-FMECA approach enables to overcome some of the limitations and pitfalls previously reported with classical FMECA approaches.

Hierarchical neural network with high storage capacity

A recent method used to optimize biased neural networks with low levels of activity is applied to a hierarchical model. As a consequence, the performance of the system is strongly enhanced. The steps to achieve optimization are analyzed in detail.

Discriminant ECOC: a heuristic method for application dependent design of error correcting output codes

We present a heuristic method for learning error correcting output codes matrices based on a hierarchical partition of the class space that maximizes a discriminative criterion. To achieve this goal, the optimal codeword separation is sacrificed in favor of a maximum class discrimination in the partitions. The creation of the hierarchical partition set is performed using a binary tree. As a result, a compact matrix with high discrimination power is obtained. Our method is validated using the UCI database and applied to a real problem, the classification of traffic sign images.

Development of a two-step screening ESI-TOF-MS method for rapid determination of significant stress-induced metabolome modifications in plant leaf extracts: the wound response in Arabidopsis thaliana as a case study.

To study the stress-induced effects caused by wounding under a new perspective, a metabolomic strategy based on HPLC-MS has been devised for the model plant Arabidopsis thaliana. To detect induced metabolites and precisely localise these compounds among the numerous constitutive metabolites, HPLC-MS analyses were performed in a two-step strategy. In a first step, rapid direct TOF-MS measurements of the crude leaf extract were performed with a ballistic gradient on a short LC-column. The HPLC-MS data were investigated by multivariate analysis as total mass spectra (TMS). Principal components analysis (PCA) and hierarchical cluster analysis (HCA) on principal coordinates were combined for data treatment. PCA and HCA demonstrated a clear clustering of plant specimens selecting the highest discriminating ions given by the complete data analysis, leading to the specific detection of discrete-induced ions (m/z values). Furthermore, pool constitution with plants of homogeneous behaviour was achieved for confirmatory analysis. In this second step, long high-resolution LC profilings on an UPLC-TOF-MS system were used on pooled samples. This allowed to precisely localise the putative biological marker induced by wounding and by specific extraction of accurate m/z values detected in the screening procedure with the TMS spectra.

Automatic Tracer-Dilution Method Used for Stage-Discharge Ratings and Streamflow Hydrographs on Small Iowa Streams, HR-285, 1990

An automatic system was designed to concurrently measure stage and discharge for the purpose of developing stage-discharge ratings and high flow hydrographs on small streams. Stage, or gage height, is recorded by an analog-to-digital recorder and discharge is determined by the constant-rate tracer-dilution method. The system measures flow above a base stage set by the user. To test the effectiveness of the system and its components, eight systems, with a variety of equipment, were installed at crest-stage gaging stations across Iowa. A fluorescent dye, rhodamine-WT, was used as the tracer. Tracer-dilution discharge measurements were made during 14 flow periods at six stations from 1986 through 1988 water years. Ratings were developed at three stations with the aid of these measurements. A loop rating was identified at one station during rapidly-changing flow conditions. Incomplete mixing and dye loss to sediment apparently were problems at some stations. Stage hydrographs were recorded for 38 flows at seven stations. Limited data on background fluorescence during high flows were also obtained.

Development of Railroad Highway Grade Crossing Consolidation Rating Formula, 2015

The goal of this project was to provide an objective methodology to support public agencies and railroads in making decisions related to consolidation of at-grade rail-highway crossings. The project team developed a weighted-index method and accompanying Microsoft Excel spreadsheet based tool to help evaluate and prioritize all public highway-rail grade crossings systematically from a possible consolidation impact perspective. Factors identified by stakeholders as critical were traffic volume, heavy-truck traffic volume, proximity to emergency medical services, proximity to schools, road system, and out-of-distance travel. Given the inherent differences between urban and rural locations, factors were considered, and weighted, differently, based on crossing location. Application of a weighted-index method allowed for all factors of interest to be included and for these factors to be ranked independently, as well as weighted according to stakeholder priorities, to create a single index. If priorities change, this approach also allows for factors and weights to be adjusted. The prioritization generated by this approach may be used to convey the need and opportunity for crossing consolidation to decision makers and stakeholders. It may also be used to quickly investigate the feasibility of a possible consolidation. Independently computed crossing risk and relative impact of consolidation may be integrated and compared to develop the most appropriate treatment strategies or alternatives for a highway-rail grade crossing. A crossing with limited- or low-consolidation impact but a high safety risk may be a prime candidate for consolidation. Similarly, a crossing with potentially high-consolidation impact as well as high risk may be an excellent candidate for crossing improvements or grade separation. The results of the highway-rail grade crossing prioritization represent a consistent and quantitative, yet preliminary, assessment. The results may serve as the foundation for more rigorous or detailed analysis and feasibility studies. Other pertinent site-specific factors, such as safety, maintenance costs, economic impacts, and location-specific access and characteristics should be considered.

Method for evaluation of root hairs of common bean genotypes

The objective of this work was to test a simple method for root hair evaluation of 21 common bean (Phaseolus vulgaris) genotypes, most of them used in breeding programs in Brazil. Hairs of basal and primary roots of 5-day old seedlings, produced on germination paper with no phosphorus addition, were visually evaluated by a rating scale after staining with 0.05% trypan blue. The method reveals variability among the genotypes, and the standard error of the mean is relatively low.

A cophenetic correlation coefficient for Tocher's method

The objective of this work was to propose a way of using the Tocher's method of clustering to obtain a matrix similar to the cophenetic one obtained for hierarchical methods, which would allow the calculation of a cophenetic correlation. To illustrate the obtention of the proposed cophenetic matrix, we used two dissimilarity matrices - one obtained with the generalized squared Mahalanobis distance and the other with the Euclidean distance - between 17 garlic cultivars, based on six morphological characters. Basically, the proposal for obtaining the cophenetic matrix was to use the average distances within and between clusters, after performing the clustering. A function in R language was proposed to compute the cophenetic matrix for Tocher's method. The empirical distribution of this correlation coefficient was briefly studied. For both dissimilarity measures, the values of cophenetic correlation obtained for the Tocher's method were higher than those obtained with the hierarchical methods (Ward's algorithm and average linkage - UPGMA). Comparisons between the clustering made with the agglomerative hierarchical methods and with the Tocher's method can be performed using a criterion in common: the correlation between matrices of original and cophenetic distances.

Cognitive Behavior Therapy for Obsessive-Compulsive Behavior in Children with Autism Spectrum Disorder: A Comparison of Hierarchical Linear Modelling and Visual Analysis

Behavioral researchers commonly use single subject designs to evaluate the effects of a given treatment. Several different methods of data analysis are used, each with their own set of methodological strengths and limitations. Visual inspection is commonly used as a method of analyzing data which assesses the variability, level, and trend both within and between conditions (Cooper, Heron, & Heward, 2007). In an attempt to quantify treatment outcomes, researchers developed two methods for analysing data called Percentage of Non-overlapping Data Points (PND) and Percentage of Data Points Exceeding the Median (PEM). The purpose of the present study is to compare and contrast the use of Hierarchical Linear Modelling (HLM), PND and PEM in single subject research. The present study used 39 behaviours, across 17 participants to compare treatment outcomes of a group cognitive behavioural therapy program, using PND, PEM, and HLM on three response classes of Obsessive Compulsive Behaviour in children with Autism Spectrum Disorder. Findings suggest that PEM and HLM complement each other and both add invaluable information to the overall treatment results. Future research should consider using both PEM and HLM when analysing single subject designs, specifically grouped data with variability.

Impact du climat organisationnel sur le façonnement des pratiques relationnelles de soin et la satisfaction professionnelle d’infirmières soignantes en milieu hospitalier

Les pratiques relationnelles de soin (PRS) sont au cœur même des normes et valeurs professionnelles qui définissent la qualité de l’exercice infirmier, mais elles sont souvent compromises par un milieu de travail défavorable. La difficulté pour les infirmières à actualiser ces PRS qui s’inscrivent dans les interactions infirmière-patient par un ensemble de comportements de caring, constitue une menace à la qualité des soins, tout en créant d’importantes frustrations pour les infirmières. En mettant l’accent sur l’aspect relationnel du processus infirmier, cette recherche, abordée sous l'angle du caring, renvoie à une vision novatrice de la qualité des soins et de l'organisation des services en visant à expliquer l’impact du climat organisationnel sur le façonnement des PRS et la satisfaction professionnelle d’infirmières soignantes en milieu hospitalier. Cette étude prend appui sur une adaptation du Quality-Caring Model© de Duffy et Hoskins (2003) qui combine le modèle d’évaluation de la qualité de Donabedian (1980, 1992) et la théorie du Human Caring de Watson (1979, 1988). Un devis mixte de type explicatif séquentiel, combinant une méthode quantitative de type corrélationnel prédictif et une méthode qualitative de type étude de cas unique avec niveaux d’analyse imbriqués, a été privilégié. Pour la section quantitative auprès d’infirmières soignantes (n = 292), différentes échelles de mesure validées, de type Likert ont permis de mesurer les variables suivantes : le climat organisationnel (global et cinq dimensions composites) ; les PRS privilégiées ; les PRS actuelles ; l’écart entre les PRS privilégiées et actuelles ; la satisfaction professionnelle. Des analyses de régression linéaire hiérarchique ont permis de répondre aux six hypothèses du volet quantitatif. Pour le volet qualitatif, les données issues des sources documentaires, des commentaires recueillis dans les questionnaires et des entrevues effectuées auprès de différents acteurs (n = 15) ont été traités de manière systématique, par analyse de contenu, afin d’expliquer les liens entre les notions d’intérêts. L’intégration des inférences quantitatives et qualitatives s’est faite selon une approche de complémentarité. Nous retenons du volet quantitatif qu’une fois les variables de contrôle prises en compte, seule une dimension composite du climat organisationnel, soit les caractéristiques de la tâche, expliquent 5 % de la variance des PRS privilégiées. Le climat organisationnel global et ses dimensions composites relatives aux caractéristiques du rôle, de l’organisation, du supérieur et de l’équipe sont de puissants facteurs explicatifs des PRS actuelles (5 % à 11 % de la variance), de l’écart entre les PRS privilégiées et actuelles (4 % à 9 %) ainsi que de la satisfaction professionnelle (13 % à 30 %) des infirmières soignantes. De plus, il a été démontré, qu’au-delà de l’important impact du climat organisationnel global et des variables de contrôle, la fréquence des PRS contribue à augmenter la satisfaction professionnelle des infirmières (ß = 0,31 ; p < 0,001), alors que l’écart entre les PRS privilégiées et actuelles contribue à la diminuer (ß = - 0,30 ; p < 0,001) dans des proportions fort similaires (respectivement 7 % et 8 %). Le volet qualitatif a permis de mettre en relief quatre ordres de facteurs qui expliquent comment le climat organisationnel façonne les PRS et la satisfaction professionnelle des infirmières. Ces facteurs sont: 1) l’intensité de la charge de travail; 2) l’approche d’équipe et la perception du rôle infirmier ; 3) la perception du supérieur et de l’organisation; 4) certaines caractéristiques propres aux patients/familles et à l’infirmière. L’analyse de ces facteurs a révélé d’intéressantes interactions dynamiques entre quatre des cinq dimensions composites du climat, suggérant ainsi qu’il soit possible d’influencer une dimension en agissant sur une autre. L’intégration des inférences quantitatives et qualitatives rend compte de l’impact prépondérant des caractéristiques du rôle sur la réalisation des PRS et la satisfaction professionnelle des infirmières, tout en suggérant d’adopter une approche systémique qui mise sur de multiples facteurs dans la mise en oeuvre d’interventions visant l’amélioration des environnements de travail infirmier en milieu hospitalier.

Ranking by rating

Each item in a given collection is characterized by a set of possible performances. A (ranking) method is a function that assigns an ordering of the items to every performance profile. Ranking by Rating consists in evaluating each item’s performance by using an exogenous rating function, and ranking items according to their performance ratings. Any such method is separable: the ordering of two items does not depend on the performances of the remaining items. We prove that every separable method must be of the ranking-by-rating type if (i) the set of possible performances is the same for all items and the method is anonymous, or (ii) the set of performances of each item is ordered and the method is monotonic. When performances are m-dimensional vectors, a separable, continuous, anonymous, monotonic, and invariant method must rank items according to a weighted geometric mean of their performances along the m dimensions.