Explanatory Factors For Variation Research Articles

Comparing the latent space of generative models

Different encodings of datapoints in the latent space of latent-vector generative models may result in more or less effective and disentangled characterizations of the different explanatory factors of variation behind the data. Many works have been recently devoted to the exploration of the latent space of specific models, mostly focused on the study of how features are disentangled and of how trajectories producing desired alterations of data in the visible space can be found. In this work we address the more general problem of comparing the latent spaces of different models, looking for transformations between them. We confined the investigation to the familiar and largely investigated case of generative models for the data manifold of human faces. The surprising, preliminary result reported in this article is that (provided models have not been taught or explicitly conceived to act differently) a simple linear mapping is enough to pass from a latent space to another while preserving most of the information. This is full of consequences for representation learning, potentially paving the way to the transformation of editing trajectories from one space to another, or the adaptation of disentanglement techniques between different generative domains.

Explanatory factors for variation in supplier sustainability performance in the automotive sector – A quantitative analysis

This article analyses differences in sustainability performances across supplier groups in the automotive industry using a multiple regression approach. The supplier sustainability performance is proxied by the self-assessment questionnaire (SAQ), a document-based assessment tool widely used in the automotive industry. The supplier groups are divided according to their headcount, business category and the region suppliers are located in. The analysis shows that the sustainability performance increases significantly with a supplier’s headcount. Furthermore, it shows that manufacturing suppliers perform significantly better compared to service providers. While there exist regional performance differences, the development stage of a country is found to be only a small significant factor influencing suppliers’ sustainability performances. Our results suggest OEM requirements in their assessments to be a major driver for supplier sustainability performance and that further analysis of regional factors is necessary.

Structural Equation Modeling to Explore Patient to Staff Ratios as an Explanatory Factor for Variation in Dialysis Facility Outcomes

Patient to staff ratios vary based on facility characteristics, and therefore have been proposed as an explanatory factor for the variation in dialysis facility outcomes. This analysis tested that hypothesis. Observational study using Dialysis Facility Report data. Reported staff numbers from the Annual Facility Survey were converted to full time equivalents (FTE). Subsequently, ratios were created for patients per FTE registered dietitian (RD), social worker, nurse, and patient care technician. Bivariate associations and structural equation modeling (SEM) were used to explore relationships between these ratios and patient outcomes: standardized mortality ratio and standardized hospitalization rate, when also considering the impact of non-modifiable facility characteristics (region, chain, profit status). Our focus was on RD staffing; therefore we also included serum phosphorus and normalized protein catabolic ratio in the model, and also conducted a sub-analysis of the 198 facilities that exceeded the KDOQI maximum of 150 patients:FTE RD. Dialysis centers in the US with at least 30 adult patients and no pediatric patients. 4035 facilities had complete data for the proposed variables. Standardized mortality ratio and standardized hospitalization rate were the primary outcomes. The mean and standard deviation for patients per FTE staff were 90.0 ± 34.0, 88.7 ± 32.8, 17.1 ± 20.5 and 11.9 ± 7.0 for RDs, social workers, nurses, and technicians, respectively. Facility characteristics impacted staffing in bivariate analyses and SEM. The only significant paths from staffing ratio to outcomes were for patient:FTE social worker to SMR (standardized beta = -0.09, 95% CI -0.13, -0.04) and Patients:FTE RD to SHR Days (standardized beta = 0.04, 95% CI 0.001, 0.09). In the sub-analysis, there were no significant paths from staffing to outcomes. This study did not provide evidence that patient per staff ratios explain variation in dialysis facility outcomes. While there are some important bivariate relationships, these disappear in more complex models. Future research should investigate the impacts of staffing ratios on individual patients, to overcome the possible ecological fallacy.

Mid-Level Feature Representation via Sparse Autoencoder for Remotely Sensed Scene Classification

Feature representation is a classic problem in the machine learning community due to the fact that different representations can entangle and hide more or less the different explanatory factors of variation behind the raw data. Especially for scene classification, its performance generally depends on the discriminative power of feature representation. Recently, unsupervised feature learning attracts tremendous attention because of its ability to learn feature representation automatically. However, reliable performance of feature representations by unsupervised learning always requires a large number of features and complex framework of mid-level feature representation. To alleviate such drawbacks, this paper presents a new framework of mid-level feature representation, which does not need learn many convolutional features during the unsupervised feature learning process, and has few parameter settings. In detail, the unsupervised feature learning method, sparse autoencoder, is employed to learn relatively small number of convolutional features from input dataset, and then extended features are extracted from the learned features by a multiple normalized difference features extraction method to compose a derivative feature set. At mid-level feature representation stage, in order to avoid poor performance of standard pooling technology in solving problems brought by rotation and translation of scene images, global feature descriptors (histogram moments, mean, variance, standard deviation) are utilized to build mid-level feature representations of images. For validation and comparison purposes, the proposed approach is evaluated via experiments with two challenging high-resolution remote sensing datasets. The results demonstrate that the approach is effective, and shows strong performance for remotely sensed scene classification.

Muscle strength does not explain standing ability in children with bilateral spastic cerebral palsy: a cross sectional descriptive study

BackgroundIn bilateral cerebral palsy (CP) muscle strength is considered important for development of gross motor functions, but its influence on standing ability has not been explored. Our aims were to examine muscle strength with respect to the ability to stand with (SwS) or without (SwoS) hand support, asymmetrical weight bearing (WB), and whether the ability to produce strength was influenced by different seated conditions.MethodsIn this cross sectional descriptive study standing posture was recorded with 3D motion analysis, and muscle strength was measured with a hand-held dynamometer, in 25 children with bilateral CP, GMFCS levels II-III, SwS (n = 14, median age 11.4 years), or SwoS, (n = 11, median age 11.4 years). Strength measurements were taken in the hip flexors, knee extensors, dorsiflexors and plantarflexors, in two seated conditions; a chair with arm- and backrests, and a stool.ResultsCompared to SwoS, children SwS stood with a more flexed posture, but presented with equal strength in the hip flexors, dorsiflexors and plantarflexors, and with somewhat more strength in the knee extensors. Despite asymmetric WB during standing, both limbs were equally strong in the two groups. No differences in strength were measured between the two seated conditions.ConclusionsDespite challenges measuring muscle strength in CP, the lower limb muscle strength cannot be considered an explanatory factor for variations in standing in this group of children with bilateral CP. The findings rather strengthen our hypothesis that deficits in the sensory systems could be as determinant for standing as muscle weakness in children with bilateral spastic CP.

Determinants of Trends in Loneliness Among Dutch Older People Over the Period 2005-2010

This study aims to investigate the influence of socio-demographic, health, and municipal characteristics on trends in loneliness among community-dwelling elderly people. Data were gathered from 4,868 and 4,773 non-institutionalized elderly people aged 65 years and above in a health survey in 2005 and 2010, respectively. Crude and adjusted multilevel models were analyzed to study the independent associations of study year and socio-demographic, health, and municipal characteristics with loneliness. Overall and across municipalities, loneliness estimates did not significantly differ between 2005 and 2010. However, among the sub-group with activity limitations, loneliness was higher in 2010 compared with 2005. This study indicates a constant trend in loneliness in the total population and across sub-groups with the exception of participants with one or more activity limitations, where loneliness increased. Individual socio-demographic and health characteristics were explanatory factors for variation in loneliness over time, whereas municipal characteristics were not.

Representation Learning: A Review and New Perspectives

The success of machine learning algorithms generally depends on data representation, and we hypothesize that this is because different representations can entangle and hide more or less the different explanatory factors of variation behind the data. Although specific domain knowledge can be used to help design representations, learning with generic priors can also be used, and the quest for AI is motivating the design of more powerful representation-learning algorithms implementing such priors. This paper reviews recent work in the area of unsupervised feature learning and deep learning, covering advances in probabilistic models, autoencoders, manifold learning, and deep networks. This motivates longer term unanswered questions about the appropriate objectives for learning good representations, for computing representations (i.e., inference), and the geometrical connections between representation learning, density estimation, and manifold learning.

Hemodynamic Correlates of Proteinuria in Chronic Kidney Disease

Brachial artery measures of BP are associated with increasing degrees of proteinuria. Whether central measures of BP or vascular stiffness are associated with increased risk of proteinuria in patients with chronic kidney disease (CKD) is unknown. Measurements of central and brachial artery BP, and aortic pulse wave velocity (PWV) were performed in a cross-sectional cohort of patients with CKD (n = 2144) from the Chronic Renal Insufficiency Cohort (CRIC) study to determine factors which predict increased risk of proteinuria. Multivariate analysis stratified by diabetes included age, ethnicity, gender, estimated glomerular filtration rate (GFR), waistline, smoking, heart rate, and medications to evaluate the relationship of hemodynamic factors and proteinuria. Brachial artery systolic BP (SBP) was important as an explanatory factor for variations in proteinuria among both diabetics (R(2) = 0.40, P < 0.0001) and non diabetics (R(2) = 0.38, P < 0.001). Measures of peripheral pulse pressure (PP), central SBP, and central pulse pressure added little to the explained variation in proteinuria beyond brachial artery SBP, whereas PWV as a measure of vascular stiffness incrementally accounted for a significant portion of variation in proteinuria beyond that explained by brachial artery SBP in diabetics (R(2) = 0.42, P < 0.001) but not non diabetics. Brachial artery SBP and PWV are both associated with variations in proteinuria in patients with CKD.

Explanatory factors for variation in the use of breast conserving surgery and radiotherapy in the Netherlands, 1990–2001

The use of Breast Conserving Surgery (BCS) followed by radiotherapy (BSC-RT) in pT1 (2cm) and pT2-tumours (2-5cm) was investigated in the Netherlands from 1990 to 2001. From the Netherlands Cancer Registry, patients were selected with invasive tumours <5.0cm. Trends were determined and explanatory factors were determined by multivariate logistic regression. Over the period 52,937 pT1-tumours and 36,285 pT2-tumours were diagnosed. The percentage BCS and BCS-RT in patients 80 years or older remained lowest. Multivariate logistic regression revealed that older age (70+), tumour size >2cm, positive clinical nodes and medium hospital size decreased the chance of BCS. For BCS-RT the same factors and negative pathological nodes decreased the chance of BCS-RT. Between regions large differences were seen. Cancer registry data are useful to monitor the guideline implementation strategies. Multidisciplinary treatment planning, surgeon and patient education could increase the use of BCS combined with RT in all age groups.

Reform Styles of Political and Administrative Elites in Majoritarian and Consensus Democracies: Public Management Reforms in New Zealand and the Netherlands

One of the most important claims advanced by comparative public management reform studies is that these reforms have been more successfully implemented in majoritarian than in consensus democracies. The claim is built on the premise that the institutional structure of a majoritarian democracy enables a parliamentary majority to implement policies unilaterally and in a desired way whereas the institutional structure of a consensus democracy forces parliamentary majorities to negotiate compromises. This claim, we argue, lacks sufficient empirical backing, is biased by studies focusing on the initiation of reforms, and needs serious rethinking in the light of recent empirical studies that downplay the salience of institutions as an explanatory factor for variations in policy reform. With The Netherlands and New Zealand as cases, this article first shows that far‐reaching reforms can be implemented in consensus systems too; secondly, it develops a model that centres on the interplay between institutions and politicians’ reform styles and applies this to these countries.

A long-term study of 370 autotransplanted premolars. Part IV. Root development subsequent to transplantation

The purpose of the present study was to analyse the extent of root development subsequent to transplantation in a material of 370 transplanted premolars. Before transplantation, the grafts were classified according to stage of root development (i.e. quarters of root length and constriction of the apical foramen) and length of tooth development (measured in mm from the cusp tip to the apical foramen). At radiographic controls the extent of root formation was registered. These figures were compared to the final tooth length of 288 non-transplanted premolars from the present material as well as to tooth length determinations of 534 premolars extracted for orthodontic purposes. Furthermore, intra-individual and inter-individual differences in tooth length were examined in a skull material with 40 pairs of premolars. These method studies indicated that root growth was probably determined genetically with great right-left accuracy and variation related to location of premolar (maxilla or mandible) and sex. A statistical analysis of the transplant material revealed that stage as well as length of tooth development assessed prior to transplantation were significantly related to final tooth length after transplantation. In most cases of transplantation performed at early stages of root development, a reduction in the final root length was seen. When a qualitative analysis of root development was performed in teeth with incomplete root formation at the time of transplantation, and where a homologous non-transplanted premolar was present, it was found that 14 per cent showed total arrest of root formation, 65 per cent partial arrest whereas 21 per cent showed no arrest (i.e. normal root formation). Premolars in ectopic position in the mandible before transplantation showed a significantly reduced root growth compared to matched pairs with normal position of the tooth graft. The present study indicated that no, partial or total damage to the Hertwig's epithelial root sheath was probably the explanatory factor for variations in root growth.