Nonparametric Bayesian Density Modeling with Gaussian Processes

We present the Gaussian process density sampler (GPDS), an exchangeable generative model for use in nonparametric Bayesian density estimation. Samples drawn from the GPDS are consistent with exact, independent samples from a distribution defined by a density that is a transformation of a function drawn from a Gaussian process prior. Our formulation allows us to infer an unknown density from data using Markov chain Monte Carlo, which gives samples from the posterior distribution over density functions and from the predictive distribution on data space. We describe two such MCMC methods. Both methods also allow inference of the hyperparameters of the Gaussian process.

Probabilistic graphical models for semi-supervised traffic classification

Traffic classification using machine learning continues to be an active research area. The majority of work in this area uses off-the-shelf machine learning tools and treats them as black-box classifiers. This approach turns all the modelling complexity into a feature selection problem. In this paper, we build a problem-specific solution to the traffic classification problem by designing a custom probabilistic graphical model. Graphical models are a modular framework to design classifiers which incorporate domain-specific knowledge. More specifically, our solution introduces semi-supervised learning which means we learn from both labelled and unlabelled traffic flows. We show that our solution performs competitively compared to previous approaches while using less data and simpler features. Copyright © 2010 ACM.

Growth of ultrahigh density vertically aligned carbon nanotube forests for interconnects.

We present a general catalyst design to synthesize ultrahigh density, aligned forests of carbon nanotubes by cyclic deposition and annealing of catalyst thin films. This leads to nanotube forests with an area density of at least 10(13) cm(-2), over 1 order of magnitude higher than existing values, and close to the limit of a fully dense forest. The technique consists of cycles of ultrathin metal film deposition, annealing, and immobilization. These ultradense forests are needed to use carbon nanotubes as vias and interconnects in integrated circuits and thermal interface materials. Further density increase to 10(14) cm(-2) by reducing nanotube diameter is possible, and it is also applicable to nanowires.

InGaN/GaN LEDs grown on Si(111): Dependence of device performance on threading dislocation density and emission wavelength

We demonstrate the growth of crack-free blue and greenemitting LED structures grown on 2-inch and 6-inch Si(111) substrates by metalorganic vapour phase epitaxy (MOVPE), using AlN nucleation layers and AlGaN buffer layers for stress management. LED device performance and its dependence on threading dislocation (TD) density and emission wavelength were studied. Despite the inherently low light extraction efficiency, an output power of 1.2 mW at 50 mA was measured from a 500 μm square planar device, emitting at 455 nm. The light output decreases dramatically as the emission wavelength increases from 455 nm to 510 nm. For LED devices emitting at similar wavelength, the light output was more than doubled when the TD density was reduced from 5×1 09 cm-2 to 2×109 cm-2. Our results clearly show that high TD density is detrimental to the overall light output, highlighting the need for further TD reduction for structures grown on Si. © 2010 Wiley-VCH Verlag GmbH & Co. KGaA.

Amorphous carbon-silicon alloys prepared by a high plasma density source

The addition of silicon to hydrogenated amorphous carbon can have the advantageous effect of lowering the compressive stress, improving the thermal stability of its hydrogen and maintaining a low friction coefficient up to high humidity. Most experiments to date have been on a-C1-xSix:H alloys deposited by RF plasma enhanced chemical vapour deposition (PECVD). This method gives alloys with considerable hydrogen content and only moderate hardness. Here, we use a high plasma density source, the electron cyclotron wave resonance (ECWR) source, to prepare films with a high deposition rate. The composition and bonding in the alloys is determined by XPS, visible and UV Raman and FTIR spectroscopy. We find that it is possible to produce hard, low stress, low friction, almost humidity insensitive a-C1-xSix:H alloys with a good optical transparency and a band gap over 2 eV.