Tree-Structured Stick Breaking Processes for Hierarchical Data

Many data are naturally modeled by an unobserved hierarchical structure. In this paper we propose a flexible nonparametric prior over unknown data hierarchies. The approach uses nested stick-breaking processes to allow for trees of unbounded width and depth, where data can live at any node and are infinitely exchangeable. One can view our model as providing infinite mixtures where the components have a dependency structure corresponding to an evolutionary diffusion down a tree. By using a stick-breaking approach, we can apply Markov chain Monte Carlo methods based on slice sampling to perform Bayesian inference and simulate from the posterior distribution on trees. We apply our method to hierarchical clustering of images and topic modeling of text data.

Nanostructured carbon films from supersonic cluster beam deposition: Structure and morphology

Nanostructured carbon thin films have been grown by deposition of cluster beams produced by a supersonic expansion. Due to separation effects typical of supersonic beams, films with different nanostructures can be grown by the simple intercepting of different regions of the cluster beam with a substrate. Films show a low-density porous structure, which has been characterized by Raman and Brillouin spectroscopy. Film morphology suggests that growth processes are similar to those occurring in a ballistic deposition regime.

Effect of size, morphology and crystallinity of seed crystal on the nucleation and growth of single grain Y-Ba-Cu-O

The effect of size, morphology and crystallinity of seed crystals on the nucleation and growth of large grain Y-Ba-Cu-O (YBCO) bulk superconductors fabricated by top seeded melt growth (TSMG) has been investigated. Seeding bulk samples with small, square shaped seed crystals leads to point nucleation and growth of the superconducting YBa2Cu3O7-y (Y-123) phase that exhibits the usual square habitual growth symmetry. The use of triangular and circular shaped seed crystals, however, modifies significantly the growth habit geometry of the grain. The use of large area seeds both increases the rate of epitaxial nucleation of the Y-123 phase and produces relatively large crystals in the incongruent melt, which decreases significantly the processing times of large grain samples. The present study is relevant to decrease processing times of samples with both preferred or no growth sectors and for multiple seeding of large grain samples which contain clean grain boundaries. © 2005 Published by Elsevier Ltd.

Decision tree-based context clustering based on cross validation and hierarchical priors

The standard, ad-hoc stopping criteria used in decision tree-based context clustering are known to be sub-optimal and require parameters to be tuned. This paper proposes a new approach for decision tree-based context clustering based on cross validation and hierarchical priors. Combination of cross validation and hierarchical priors within decision tree-based context clustering offers better model selection and more robust parameter estimation than conventional approaches, with no tuning parameters. Experimental results on HMM-based speech synthesis show that the proposed approach achieved significant improvements in naturalness of synthesized speech over the conventional approaches. © 2011 IEEE.