Pitman-Yor Diffusion Trees

We introduce the Pitman Yor Diffusion Tree (PYDT) for hierarchical clustering, a generalization of the Dirichlet Diffusion Tree (Neal, 2001) which removes the restriction to binary branching structure. The generative process is described and shown to result in an exchangeable distribution over data points. We prove some theoretical properties of the model and then present two inference methods: a collapsed MCMC sampler which allows us to model uncertainty over tree structures, and a computationally efficient greedy Bayesian EM search algorithm. Both algorithms use message passing on the tree structure. The utility of the model and algorithms is demonstrated on synthetic and real world data, both continuous and binary.

Phase separation induced by Au catalysts in ternary InGaAs nanowires.

We report a novel phase separation phenomenon observed in the growth of ternary In(x)Ga(1-x)As nanowires by metalorganic chemical vapor deposition. A spontaneous formation of core-shell nanowires is investigated by cross-sectional transmission electron microscopy, revealing the compositional complexity within the ternary nanowires. It has been found that for In(x)Ga(1-x)As nanowires high precursor flow rates generate ternary In(x)Ga(1-x)As cores with In-rich shells, while low precursor flow rates produce binary GaAs cores with ternary In(x)Ga(1-x)As shells. First-principle calculations combined with thermodynamic considerations suggest that this phenomenon is due to competitive alloying of different group-III elements with Au catalysts, and variations in elemental concentrations of group-III materials in the catalyst under different precursor flow rates. This study shows that precursor flow rates are critical factors for manipulating Au catalysts to produce nanowires of desired composition.

Influence of nanowire density on the shape and optical properties of ternary InGaAs nanowires.

We have synthesized ternary InGaAs nanowires on (111)B GaAs surfaces by metal-organic chemical vapor deposition. Au colloidal nanoparticles were employed to catalyze nanowire growth. We observed the strong influence of nanowire density on nanowire height, tapering, and base shape specific to the nanowires with high In composition. This dependency was attributed to the large difference of diffusion length on (111)B surfaces between In and Ga reaction species, with In being the more mobile species. Energy dispersive X-ray spectroscopy analysis together with high-resolution electron microscopy study of individual InGaAs nanowires shows large In/Ga compositional variation along the nanowire supporting the present diffusion model. Photoluminescence spectra exhibit a red shift with decreasing nanowire density due to the higher degree of In incorporation in more sparsely distributed InGaAs nanowires.

MoT - Mixture of trees probabilistic graphical model for video segmentation

We present a novel mixture of trees (MoT) graphical model for video segmentation. Each component in this mixture represents a tree structured temporal linkage between super-pixels from the first to the last frame of a video sequence. Our time-series model explicitly captures the uncertainty in temporal linkage between adjacent frames which improves segmentation accuracy. We provide a variational inference scheme for this model to estimate super-pixel labels and their confidences in nearly realtime. The efficacy of our approach is demonstrated via quantitative comparisons on the challenging SegTrack joint segmentation and tracking dataset [23].