Feature selection in the reconstruction of complex network representations of spectral data

Complex networks have been extensively used in the last decade to characterize and analyze complex systems, and they have been recently proposed as a novel instrument for the analysis of spectra extracted from biological samples. Yet, the high number of measurements composing spectra, and the consequent high computational cost, make a direct network analysis unfeasible. We here present a comparative analysis of three customary feature selection algorithms, including the binning of spectral data and the use of information theory metrics. Such algorithms are compared by assessing the score obtained in a classification task, where healthy subjects and people suffering from different types of cancers should be discriminated. Results indicate that a feature selection strategy based on Mutual Information outperforms the more classical data binning, while allowing a reduction of the dimensionality of the data set in two orders of magnitude

Influence of particle morphology on nanostructural feature and conducting property in Sm-doped CeO2 sintered body

Doped ceria (CeO2,) compounds are fluorite type oxides, which show oxide ionic conductivity higher than yttria stabilized zirconia (YSZ), in oxidizing atmospheres. As a consequence of this, considerable interest has been shown in application of these materials for 'low (500-650 degreesC)' or 'intermediate (650-800 degreesC)' temperature operation, solid oxide fuel cells (SOFCs). In this study, the authors prepared two kinds of nanosize Sm-doped CeO2 particles with different morphologies: one type was round and the other was elongated. Processing these powders with different morphology produced dense materials with very different ionic conducting properties and different nanoscale microstructures. Since both particles are very fine and well dispersed, sintered bodies with high density (relative density >95% of theoretical) could be prepared using both types of powder particles. The electrical conductivity of sintered bodies prepared from these powders with different starting morphologies was very different. Materials prepared from particles having a round shape were much higher than those produced using powders with an elongated morphology. Measured activation energies of the corresponding sintered samples showed a similar trend; round particles (60 kJ/mol), elongated particles (74 kJ/mol). While X-ray diffraction (XRD) profiles of these sintered materials were identical, diffuse scatter was observed in the back.-round of selected area electron diffraction pattern recorded from both sintered bodies. This indicated an underlying structure that appeared to have been influenced by the processing technology. Detailed observation using high-resolution transmission electron microscopy (HR-TEM) revealed that the size of microdomain with ordering of cations in the sintered body made from round shape particles was much smaller than that of the sintered body made from elongated particles. Accordingly, it is concluded that the morphology of doped CeO2 powders strongly influenced the microdomain size and electrolytic properties in the doped CeO2 sintered body. (C) 2004 Elsevier B.V. All rights reserved.

Visualisation of heterogeneous data with simultaneous feature saliency using Generalised Generative Topographic Mapping

Most machine-learning algorithms are designed for datasets with features of a single type whereas very little attention has been given to datasets with mixed-type features. We recently proposed a model to handle mixed types with a probabilistic latent variable formalism. This proposed model describes the data by type-specific distributions that are conditionally independent given the latent space and is called generalised generative topographic mapping (GGTM). It has often been observed that visualisations of high-dimensional datasets can be poor in the presence of noisy features. In this paper we therefore propose to extend the GGTM to estimate feature saliency values (GGTMFS) as an integrated part of the parameter learning process with an expectation-maximisation (EM) algorithm. The efficacy of the proposed GGTMFS model is demonstrated both for synthetic and real datasets.

Lady Nugent and Tom Cringle on the veranda : early nineteenth century observations on a Caribbean architectural feature

This study of the veranda as seen through the eyes of Lady Maria Nugent and Michael Scott, alias Tom Cringle, clearly demonstrates the important role that the piazza, as it was then more commonly known, played in the life of early nineteenth century Caribbean colonial society. The popularity of the veranda throughout the region, in places influenced by different European as well as African cultures, and among all classes of people, suggests that the appeal of this typical feature was based on something more than architectural fashion. A place of relative comfort in hot weather, the veranda is also a space at the interface of indoors and outdoors which allows for a wide variety of uses, for solitary or small or large group activities, many of which were noted by Nugent and Scott. Quintessentially, the veranda is a place in which to relax and take pleasure, not least of which is the enjoyment of the prospect, be it a panoramic view, a peaceful garden or a lively street scene. Despite the great changes in the nature of society, in the Caribbean and in many other parts of the world, the veranda and related structures such as the balcony continue to play at least as important a role in daily life as they did two centuries ago. The veranda of today’s Californian or Australian bungalow, and the balcony of the apartment block in the residential area of the modern city are among the contemporary equivalents of the lower and upper piazzas of Lady Nugent’s and Tom Cringle’s day.

2D-3D Face Recognition Based on PCA and Feature Modelling

Hybrid face recognition, using image (2D) and structural (3D) information, has explored the fusion of Nearest Neighbour classifiers. This paper examines the effectiveness of feature modelling for each individual modality, 2D and 3D. Furthermore, it is demonstrated that the fusion of feature modelling techniques for the 2D and 3D modalities yields performance improvements over the individual classifiers. By fusing the feature modelling classifiers for each modality with equal weights the average Equal Error Rate improves from 12.60% for the 2D classifier and 12.10% for the 3D classifier to 7.38% for the Hybrid 2D+3D clasiffier.