An efficient feature selection method for mobile devices with application to activity recognition

This paper presents a feature selection method for data classification, which combines a model-based variable selection technique and a fast two-stage subset selection algorithm. The relationship between a specified (and complete) set of candidate features and the class label is modelled using a non-linear full regression model which is linear-in-the-parameters. The performance of a sub-model measured by the sum of the squared-errors (SSE) is used to score the informativeness of the subset of features involved in the sub-model. The two-stage subset selection algorithm approaches a solution sub-model with the SSE being locally minimized. The features involved in the solution sub-model are selected as inputs to support vector machines (SVMs) for classification. The memory requirement of this algorithm is independent of the number of training patterns. This property makes this method suitable for applications executed in mobile devices where physical RAM memory is very limited. An application was developed for activity recognition, which implements the proposed feature selection algorithm and an SVM training procedure. Experiments are carried out with the application running on a PDA for human activity recognition using accelerometer data. A comparison with an information gain based feature selection method demonstrates the effectiveness and efficiency of the proposed algorithm.

Computer simulating a trial of a load-bearing implant: example of an intramedullary prosthesis

Computational modelling is becoming ever more important for obtaining regulatory approval for new medical devices. An accepted approach is to infer performance in a population from an analysis conducted for an idealised or ‘average’ patient; we present here a method for predicting the performance of an orthopaedic implant when released into a population—effectively simulating a clinical trial. Specifically we hypothesise that an analysis based on a method for predicting the performance in a population will lead to different conclusions than an analysis based on an idealised or ‘average’ patient. To test this hypothesis we use a finite element model of an intramedullary implant in a bone whose size and remodelling activity is different for each individual in the population. We compare the performance of a low Young’s modulus implant (View the MathML source) to one with a higher Young’s modulus (200 GPa). Cyclic loading is applied and failure is assumed when the migration of the implant relative to the bone exceeds a threshold magnitude. The analysis for an idealised of ‘average’ patient predicts that the lower modulus device survives longer whereas the analysis simulating a clinical trial predicts no statistically-significant tendency (p=0.77) for the low modulus device to perform better. It is concluded that population-based simulations of implant performance–simulating a clinical trial–present a very valuable opportunity for more realistic computational pre-clinical testing of medical devices.

Comparing large graphs efficiently by margins of feature vectors

Measuring the structural similarity of graphs is a challenging and outstanding problem. Most of the classical approaches of the so-called exact graph matching methods are based on graph or subgraph isomorphic relations of the underlying graphs. In contrast to these methods in this paper we introduce a novel approach to measure the structural similarity of directed and undirected graphs that is mainly based on margins of feature vectors representing graphs. We introduce novel graph similarity and dissimilarity measures, provide some properties and analyze their algorithmic complexity. We find that the computational complexity of our measures is polynomial in the graph size and, hence, significantly better than classical methods from, e.g. exact graph matching which are NP-complete. Numerically, we provide some examples of our measure and compare the results with the well-known graph edit distance. (c) 2006 Elsevier Inc. All rights reserved.

Study of structural change in Wyodak coal in high pressure CO2 by small-angle neutron scattering

Small angle neutron scattering (SANS) has been applied to examine the effect of high pressure CO2 on the structure of Wyodak coal. Significant decrease in the scattering intensities upon exposure of the coal to high pressure CO2 showed that high pressure CO2 rapidly adsorbs on the coal and reaches to all pores in the structure. This is confirmed by strong and steep exothermic peaks observed on DSC scans during coal/ CO2 interactions. In situ small angle neutron scattering on coal at high pressure CO2 atmosphere showed an increase in scattering intensities with time suggesting that after adsorption, high pressure CO2 immediately begins to diffuse into the coal matrix, changes the macromolecular structure of the coal, swells the matrix and probably creates microporosity in coal structure by extraction of volatile components from coal. Significant decrease in the glass transition temperature of coal caused by high pressure CO2 also confirms that CO2 at elevated pressures dissolve in the coal matrix, results in significant plasticization and physical rearrangement of the coal’s macromolecular structure.

Multiaxial Deformation of Polyethylene and Polyethylene/Clay Nanocomposites: In Situ Synchrotron Small Angle and Wide Angle X-Ray Scattering Study

A unique in situ multiaxial deformation device has been designed and built specifically for simultaneous synchrotron small angle X-ray scattering (SAXS) and wide angle X-ray scattering (WAXS) measurements. SAXS and WAXS patterns of high-density polyethylene (HDPE) and HDPE/clay nanocomposites were measured in real time during in situ multiaxial deformation at room temperature and at 55 degrees C. It was observed that the morphological evolution of polyethylene is affected by the existence of clay platelets as well as the deformation temperature and strain rate. Martensitic transformation of orthorhombic into monoclinic crystal phases was observed under strain in HDPE, which is delayed and hindered in the presence of clay nanoplatelets. From the SAXS measurements, it was observed that the thickness of the interlamellar amorphous region increased with increasing strain, which is due to elongation of the amorphous chains. The increase in amorphous layer thickness is slightly higher for the nanocomposites compared to the neat polymer. (C) 2011 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 49: 669-677, 2011

Frequency selective surface using nested split ring slot elements as a lens with mechanically reconfigurable beam steering capability

The design is described of a double layer frequency selective surface which can produce a differential phase shift of 180 ° as the wave propagates through it at normal incidence. The hand of an applied circularly polarized signal is reversed due to the 180° phase shift, and it is demonstrated that the exit circularly polarized output signal can be phase advanced or phase retarded by 180 ° upon rotation of the elements comprising the structure. This feature allows the surface to act as a spatial phase shifter. In this paper the beam steering capabilities of a 10 × 10 array of such elements are demonstrated. Here the individual elements comprising the array are rotated relative to each other in order to generate a progressive phase shift. At normal incidence the 3 dB Axial Ratio Bandwidth for LHCP to RHCP conversion is 5.3% and the insertion loss was found to be -2.3 dB, with minimum axial ratio of 0.05 dB. This array is shown to be able to steer a beam from -40 ° to +40 ° while holding axial ratio at the pointing angle to below 4 dB. The measured radiation patterns match the theoretical calculation and full-wave simulation results. © 2010 IEEE.