Prediction- and simulation-error based perceptron training: Solution space analysis and a novel combined training scheme

Previous papers have noted the difficulty in obtaining neural models which are stable under simulation when trained using prediction-error-based methods. Here the differences between series-parallel and parallel identification structures for training neural models are investigated. The effect of the error surface shape on training convergence and simulation performance is analysed using a standard algorithm operating in both training modes. A combined series-parallel/parallel training scheme is proposed, aiming to provide a more effective means of obtaining accurate neural simulation models. Simulation examples show the combined scheme is advantageous in circumstances where the solution space is known or suspected to be complex. (c) 2006 Elsevier B.V. All rights reserved.

Rapid forensic analysis and identification of "lilac" architectural finishes using Raman spectroscopy

The potential of Raman spectroscopy to discriminate between architectural finishes (household paint) has been investigated using a test set of 51

SO and CS observations of molecular clouds - II. Analysis and modelling of the abundance ratios - probing O-2/CO with SO/CS?

We here analyse the observational SO and CS data presented in Nilsson ct al. (2000). The SO/CS integrated intensity ratio maps are presented for 19 molecular clouds, together with tables of relevant ratios at strategic positions, where we have also observed (SO)-S-34 and/or (CS)-S-34. The SO/CS abundance ratio as calculated from an LTE analysis is highly varying within and between the sources. Our isotopomer observations and Monte Carlo simulations verify that this is not an artifact due to optical depth problems. The variation of the maximum SO/CS abundance ratio between the clouds is 0.2-7. The largest variations within a cloud are found for the most nearby objects, possibly indicating resolution effects. We have also performed time dependent chemical simulations. We compare the simulations with our observed SO/CS abundance ratios and suggest a varying oxygen to carbon initial abundance, differing temporal evolution, density differences and X-ray sources associated with young stellar objects as possible explanations to the variations. In particular, the observed variation of the maximum SO/CS abundance ratio between the clouds can be explained by using initial O/C+ abundance ratios in the range 1.3-2.5. We finally derive a relationship between the SO/CS and O-2/CO abundance ratios, which may be used as a guide to find the most promising interstellar O-2 search targets.

A time-domain approach to the analysis and modeling of on-body propagation characteristics using synchronized measurements at 2.45 GHz

Modeling of on-body propagation channels is of paramount importance to those wishing to evaluate radio channel performance for wearable devices in body area networks (BANs). Difficulties in modeling arise due to the highly variable channel conditions related to changes in the user's state and local environment. This study characterizes these influences by using time-series analysis to examine and model signal characteristics for on-body radio channels in user stationary and mobile scenarios in four different locations: anechoic chamber, open office area, hallway, and outdoor environment. Autocorrelation and cross-correlation functions are reported and shown to be dependent on body state and surroundings. Autoregressive (AR) transfer functions are used to perform time-series analysis and develop models for fading in various on-body links. Due to the non-Gaussian nature of the logarithmically transformed observed signal envelope in the majority of mobile user states, a simple method for reproducing the failing based on lognormal and Nakagami statistics is proposed. The validity of the AR models is evaluated using hypothesis testing, which is based on the Ljung-Box statistic, and the estimated distributional parameters of the simulator output compared with those from experimental results.

Analysis and comparison of displacement granular pile anchors

Granular piles can resist only compressive and shear loads owing to their inherent nature. By a simple modification of providing a pedestal/geogrid at the bottom and attaching a cable to the same, they are made to resist pullout/uplift forces. This paper presents an analysis of granular pile anchor (GPA), considering it and the in situ soil to behave linearly and the in situ ground to be semi-infinite. A parametric study presents results in the form of variations of normalised shear stress, displacement influence coefficient and axial uplift force with depth with relative stiffness factor. Two methods for the estimation of deformation moduli of the GPA and the in situ soil are proposed. Based on the estimated values of the moduli, the displacements of GPA were estimated and the results compared with test results of Kumar (2002). The predicted displacements compare well with the measured ones.

Developing a model of innovation implementation for UK SMEs: A path analysis and explanatory case analysis

Continuous large-scale changes in technology and the globalization of markets have resulted in the need for many SMEs to use innovation as a means of seeking competitive advantage where innovation includes both technological and organizational perspectives (Tapscott, 2009). However, there is a paucity of systematic and empirical research relating to the implementation of innovation management in the context of SMEs. The aim of this article is to redress this imbalance via an empirical study created to develop and test a model of innovation implementation in SMEs. This study uses Structural Equation Modelling (SEM) to test the plausibility of an innovation model, developed from earlier studies, as the basis of a questionnaire survey of 395 SMEs in the UK. The resultant model and construct relationship results are further probed using an explanatory multiple case analysis to explore ‘how’ and ‘why’ type questions within the model and construct relationships. The findings show that the
effects of leadership, people and culture on innovation implementation are mediated by business improvement activities relating to Total Quality Management/Continuous Improvement (TQM/CI) and product and process developments. It is concluded that SMEs have an opportunity to leverage existing quality and process improvement activities to move beyond continuous
improvement outcomes towards effective innovation implementation. The article concludes by suggesting areas suitable for further research.