Generalized Loopy 2U: A New Algorithm for Approximate Inference in Credal Networks

Credal nets generalize Bayesian nets by relaxing the requirement of precision of probabilities. Credal nets are considerably more expressive than Bayesian nets, but this makes belief updating NP-hard even on polytrees. We develop a new efficient algorithm for approximate belief updating in credal nets. The algorithm is based on an important representation result we prove for general credal nets: that any credal net can be equivalently reformulated as a credal net with binary variables; moreover, the transformation, which is considerably more complex than in the Bayesian case, can be implemented in polynomial time. The equivalent binary credal net is updated by L2U, a loopy approximate algorithm for binary credal nets. Thus, we generalize L2U to non-binary credal nets, obtaining an accurate and scalable algorithm for the general case, which is approximate only because of its loopy nature. The accuracy of the inferences is evaluated by empirical tests.

Melt Processing and Properties of Polyamide 6/Graphene Nanoplatelet Composites

In this paper, the processing and characterization of Polyamide 6 (PA6) / graphite nanoplatelets
(GNPs) composites is reported. PA6/GNPs composites were prepared by melt-mixing using an
industrial, co-rotating, intermeshing, twin-screw extruder. A bespoke screw configuration was used
that was designed in-house to enhance nanoparticle dispersion into a polymer matrix. The effects of
GNPs type (xGnP® M-5 and xGnP® C-500), GNPs content, and extruder screw speed on the bulk
properties of the PA6/GNPs nanocomposites were investigated. Results show a considerable
improvement in the thermal and mechanical properties of PA6/GNPs composites, as compared with
the unfilled PA6 polymer. An increase in crystallinity (%Xc) with increasing GNPs content, and a
change in shape of the crystallization exotherms (broadening) and melting endotherms, both suggest a
change in the crystal type and perfection. An increase in tensile modulus of as much as 376% and
412% was observed for PA6/M-5 xGnP® and PA6/C-500 xGnP® composites, respectively, at filler
contents of 20wt%. The enhancement of Young’s modulus and yield stress can be attributed to the
reinforcing effect of GNPs and their uniform dispersion in the PA6 matrix. The rheological response
of the composite resembles that of a ‘pseudo-solid’, rather than a molten liquid, and analysis of the
rheological data indicates that a percolation threshold was reached at GNPs contents of between 10–
15wt%. The electrical conductivity of the composite also increased with increasing GNPs content,
with an addition of 15wt% GNPs resulting in a 6 order-of-magnitude increase in conductivity. The
electrical percolation thresholds of all composites were between 10–15wt%.

Micromechanical Modelling of Non-Homogenous Materials by Meshless Methods

A detailed study of bi-material composites, using meshless methods (MMs), is presented in this paper. Firstly, representative volume elements (RVEs) for different bi-material combinations are analysed by the element-free Galerkin (EFG) method in order to confirm the effective properties of heterogeneous material through homogenization. The results are shown to be in good agreement with experimental results and those obtained using the finite element method (FEM) which required a higher node density. Secondly, a functionally graded material (FGM), with a crack, is analysed using the EFG method. This investigation was motivated by the possibility of replacing the distinct fibrematrix interface with a FGM interface. Finally, an illustrative example showing crack propagation, in a two-dimension micro-scale model of a SiC/Al composite is presented. 

A systematic review of population based epidemiological studies in Myasthenia Gravis

Background: The aim was to collate all myasthenia gravis (MG) epidemiological studies including AChR MG and MuSK MG specific studies. To synthesize data on incidence rate (IR), prevalence rate (PR) and mortality rate (MR) of the condition and investigate the influence of environmental and technical factors on any trends or variation observed.

Methods: Studies were identified using multiple sources and meta-analysis performed to calculate pooled estimates for IR, PR and MR.

Results: 55 studies performed between 1950 and 2007 were included, representing 1.7 billion population-years. For All MG estimated pooled IR (eIR): 5.3 per million person-years (C.I.: 4.4, 6.1), range: 1.7 to 21.3; estimated pooled PR: 77.7 per illion persons (C.I.: 64.0, 94.3), range 15 to 179; MR range 0.1 to 0.9 per millions person-years. AChR MG eIR: 7.3 (C.I.: 5.5, 7.8), range: 4.3 to 18.0; MuSK MG IR range: 0.1 to 0.32. However marked variation persisted between populations studied with similar methodology and in similar areas.

Conclusions: We report marked variation in observed frequencies of MG. We show evidence of increasing frequency of MG with year of study and improved study quality. This probably reflects improved case ascertainment. But other factors must also influence disease onset resulting in the observed variation in IR across geographically and genetically similar populations.

Atmospheric Pressure DC Plasma Processing of TIO2/PEDOT:PSS Nanocomposites

In this work we demonstrate the synthesis of a TiO2/PEDOT:PSS nanocomposite material in aqueous solution through atmospheric pressure direct current (DC) plasma processing at room temperature. The dispersion of the TiO2 nanoparticles is enhanced after microplasma processing, and TiO2/polymer hybrid nanoparticles with a distinct core shell structure have been obtained. We have observed increased TiO2/PEDOT:PSS nanocomposite electrical conductivity due to microplasma processing. The improvement in nanocomposite properties is due to the enhanced dispersion and stability in liquid polymer of microplasma treated TiO2 nanoparticles. Both plasma induced surface charge and nanoparticle surface termination with specific plasma chemical species are thought to provide an enhanced barrier to nanoparticle agglomeration and promote nanoparticle-polymer bonding, which is expected to have a significant benefit in materials processing with inorganic nanoparticles for wide range of applications.