Boosted multivariate trees for longitudinal data

Machine learning methods provide a powerful approach for analyzing longitudinal data in which repeated measurements are observed for a subject over time. We boost multivariate trees to fit a novel flexible semi-nonparametric marginal model for longitudinal data. In this model, features are assumed to be nonparametric, while feature-time interactions are modeled semi-nonparametrically utilizing P-splines with estimated smoothing parameter. In order to avoid overfitting, we describe a relatively simple in sample cross-validation method which can be used to estimate the optimal boosting iteration and which has the surprising added benefit of stabilizing certain parameter estimates. Our new multivariate tree boosting method is shown to be highly flexible, robust to covariance misspecification and unbalanced designs, and resistant to overfitting in high dimensions. Feature selection can be used to identify important features and feature-time interactions. An application to longitudinal data of forced 1-second lung expiratory volume (FEV1) for lung transplant patients identifies an important feature-time interaction and illustrates the ease with which our method can find complex relationships in longitudinal data.     

Efficient compression and network adaptive video coding for distributed video surveillance

The availability of cheap network based video cameras and the prevalence of wireless networks has lead to a major thrust towards deployment of large scale Distributed Video Surveillance (DVS) systems. This has opened up an important area of research to deal with the issues involved in DVS system for efficient collection and transmission of large scale video streams from the cameras at the guarded sites, to the end users in possibly constrained network conditions. In this paper, we propose a framework based on content-based video classification and scalable compression scheme to provide a robust bandwidth efficient video transmission for DVS. The scheme builds on a Discrete Wavelet Transform (DWT) based Color-Set Partitioning for Hierarchical Trees (CSPIHT) coding to obtain a scalable bitstream. Wavelet domain segmentation and compression assists in development of a DVS architecture. The architecture includes a novel module for dynamic allocation of Network bandwidth based on the current available resources and constraints. Different frame constituents are optimally coded based on their relative significance, perceptual quality, and available estimate of network bandwidth. Experimental result over different video sequences and simulations for Network conditions demonstrate the efficient performance of the approach.     

On the relationship between the electronic density of states and the superconducting critical temperature of disordered V3Si

The available data on the effect of neutron and α-particle irradiation on the superconducting properties and the low-temperature specific heat capacity of single- and polycrystalline V₃Si were analyzed in a unified manner. It was found that the electron-phonon renormalization parameter λ does not vary simply proportionally to the bare density of electron states at the Fermi level as had been earlier conjectured by Dynes and Varma, nor does it vary in exactly the same manner as found by Fradin and Williamson to hold for a variety of vanadium-based A15 structure compounds and pseudobinary alloys. When the behavior of V₃Si is compared with that of Nb₃Sn, Nb₃Al, and Nb₃Ge, it is seen that the response to disorder of parameters of importance in the superconductivity of these compounds does not follow any simple, universal behavior.     

High-gain, V-band, low-noise MMIC amplifiers usingpseudomorphic MODFETs

State-of-the-art, 60-GHz, low-noise MMICs based on pseudomorphic modulation-doped FETs, with 0.25-μm×60-μm gates offset 0.3 μm from the source ohmic, are discussed. Single-state low-noise amplifiers (LNAs) exhibited minimum noise figures of 2.90 dB with 4.1 dB of associated gain at 59.25 GHz. Dual-state MMICs had minimum noise figures of 3.5 dB and 10.8 dB of associated gain at 58.50 GHz. Cascaded four-stage LNAs (two dual-stage MMICs) had minimum noise figures of 3.7 dB and over 20.7 dB of associated gain at 58.0 GHz. Finally, when biased for maximum gain, the four-stage amplifier exhibited over 30.4 dB of gain at 60.0 GHz     

Bench-scale evaluations and mechanistic studies of intumescent/fire-protective coatings for polyvinyl chloride nitrile rubber

The object of this study was to identify, for future large-scale testing, and adequate fire-protective coating for the substrate, polyvinyl chloride (PVC) nitrile rubber (MIL-P-15280 Rev H). Fifteen fire protective coatings which included ten intumescent, two ablatives, two fire retardants and one photon diffusive coatings were evaluated using bench-scale thermal insults. To discern the possible mechanism(s) whereby the coatings offer fire protection, the effect of radiant heat versus the flame source employed was investigated and thermogravimetric analysis (TGA) of the neat coatings was performed. In selected cases, the effect of coating thickness and repeatability of the results were evaluated. The differences in coating effectiveness, and the varied repeatability in three of the four cases examined, focus on the inherent complexity of intumescence, as well as on the importance of scaled-up testing of coatings that appeared promising. Thermogravimetric analyses of the coatings indicate that two selected TGA measurements/parameters, when used in conjunction, appear useful in characterizing mechanistic differences between the less effective and more effective coatings. Based on the rigorous criterion adopted, one candidate (a water-based intumescent coating at 100 mil (2.54 mm) thickness) appears promising for the substrate, PVC nitrile rubber. Its effectiveness is attributable to it exhibiting good insulative and re-radiative properties and an apparent low susceptibility to crack formation.     

Long-term synaptic plasticity in the honeybee

A monosynaptic response was recorded in vivo in the mushroom body of the bee brain, an important site for memory consolidation. Focal electrical stimulation of a major afferent input evoked an extracellular field potential that consisted of a presynaptic fiber volley and a postsynaptic response. We report a long-lasting potentiation of the synaptic response (2.6-fold increase; < or = 3.5 h). Potentiation of the response was induced by low-frequency stimulation (0.02-1.0 Hz), was input specific, and was maintained in the absence of stimulation. Paired-pulse facilitation of the response was converted to paired-pulse depression after potentiation, suggesting a presynaptic mechanism. This is the first demonstration of long-term synaptic plasticity in the insect brain.     

Polymorphism of pericentromere heterochromatin of polytene chromosomes of ovarian trophocytes in natural populations of the malaria mosquito Anopheles messeae Fall

The HBsAg status and demographic data of 2480 pregnant women who attended antenatal clinics at Maternal and Child Health Centres in Hong Kong were collected by means of a self-administered questionnaire over a 1-week period in July 1996, to explore the underlying reason of a higher than expected HBsAg prevalence. Local women constituted 49.2% of the sample, mainland Chinese 39.7% and others 11.1%. The overall HBsAg prevalence was 10.0%. When related to place of birth, those born in Hong Kong had a prevalence of 8.4% whereas the prevalence of those born in mainland China was 13.1% (P < 0.001). The overall HBsAg carriage rate is high because of a higher rate in immigrants in the community. It is apparent that the HBsAg prevalence of local people in Hong Kong has been decreasing in the past decade. Overall, the current HBsAg carriage rate in the local adult population is estimated to have declined to about 8%.     

Cortical variations of the dimension of correlation in children with attention deficit hyperactivity syndrome

INTRODUCTION AND OBJECTIVE: The syndrome of deficient attention and hyperactivity affects approximately 5% of children of school age. Clinically it is characterized by deficient attention, impulsiveness and excessive motor activity. Developments in nonlinear dynamic theory have made it possible to think of modelling cerebral activity as a dynamic system of chaotic type. One the most widely standardized measurements is the dimension of correlation (D2). In this paper we study the electroencephalographic traces of a control group of 9 healthy children and another group of 19 children with the syndrome of deficient attention, at rest and whilst carrying out a visuomotor task (Test of perception of differences). MATERIAL AND METHODS: The dimension of correlation was estimated for both groups. Statistical comparison was made between the cortical distributions of the D2 between both types of recordings and between both groups. RESULTS: The children with lack of attention showed a larger number of activated cortical areas than those of the control group when carrying out the same task. However, during the resting state there were no differences in the D2 between the control children and the children with deficient attention. The indicates that both groups have the same basal level of activation. CONCLUSION: The dimension of correlation is a method which permits the demonstration of an increase in neuronal activity in the stimulated areas, in this case in the occipital region.