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.

Value roadmapping: A structured approach for early stage technology investment decisions

Deciding to invest in early stage technologies is one of the most important tasks of technology management and arguably also the most uncertain. It assumes a particular significance in the rise of technology companies in emerging economies, which have to make appropriate investment decisions. Technology managers already have a wide range of methods and tools at their disposal, but these are mostly focussed on quantitative measures such as discounted cash flow and real options techniques. However, in the early stages of technology development there seems to be a lot of dissatisfaction with these techniques as there appears to be a lack of accuracy with respect to the underlying assumptions that these models require. In order to complement these models this paper will discuss an alternative approach that we call value road-mapping. By adapting roadmapping techniques the potential value streams of early stages technologies can be plotted and hence a clearer consensus based picture of the future potential of new technologies emerges. Roadmapping is a workshop-based process bringing together multifunctional perspectives, and supporting communication in particular between technical and commercial groups. The study is work in progress and is based on a growing number of cases. (c) 2006 PICMET.

Electric conduction improvement of well-structured multi-walled carbon nanotubes

We characterized the electrical conductance of well-structured multi-walled carbon nanotubes (MWCNTs) which had post-treated by a rapid vacuum arc thermal annealing process and structure defects in these nanotubes are removed. We found that the after rapid vacuum arc annealing, the conductivity of well-structured MWCNTs can be improved by an order of magnitude. We also investigated the conductivity of MWCNTs bundle by the variation of temperatures. These results show that the conductance of annealed defect-free MWCNTs is sensitive to temperature imply the phonon scatting dominated the electron conductions. Compare to the well-structured MWCNTs, the defect scattering dominated the electron conduction in the as-grown control sample which has large amount of structure defects. A detail measurement of electron conduction from an individual well-structured MWCNT shows that the conductivity increases with temperatures which imply such MWCNTs exhibited semiconductor properties. We also produced back-gated field-effect transistors using these MWCNTs. It shows that the well-structured MWCNT can act as p-type semiconductor. © 2010 IEEE.