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.

Support vector machine classification for large data sets via minimum enclosing ball clustering

Support vector machine (SVM) is a powerful technique for data classification. Despite of its good theoretic foundations and high classification accuracy, normal SVM is not suitable for classification of large data sets, because the training complexity of SVM is highly dependent on the size of data set. This paper presents a novel SVM classification approach for large data sets by using minimum enclosing ball clustering. After the training data are partitioned by the proposed clustering method, the centers of the clusters are used for the first time SVM classification. Then we use the clusters whose centers are support vectors or those clusters which have different classes to perform the second time SVM classification. In this stage most data are removed. Several experimental results show that the approach proposed in this paper has good classification accuracy compared with classic SVM while the training is significantly faster than several other SVM classifiers.

Psychological problems in children with cerebral palsy: A cross-sectional European study

Objectives: To describe psychological symptoms in 8–12-year-old children with cerebral palsy; to investigate predictors of these symptoms and their impact on the child and family.

Design: A cross-sectional multi-centre survey.

Participants: Eight hundred and eighteen children with cerebral palsy, aged 8–12 years, identified from population-based registers of cerebral palsy in eight European regions and from multiple sources in one further region.

Main outcome measures: The Strengths and Difficulties Questionnaire (SDQ)P4-16 and the Total Difficulties Score (TDS) dichotomised into normal/borderline (TDS = 16) versus abnormal (TDS > 16).

Statistical analysis: Multilevel, multivariable logistic regression to relate the presence of psychological symptoms to child and family characteristics.

Results: About a quarter of the children had TDS > 16 indicating significant psychological symptoms, most commonly in the domain Peer Problems. Better gross motor function, poorer intellect, more pain, having a disabled or ill sibling and living in a town were independently associated with TDS > 16. The risk of TDS > 16 was odds ratio (OR) = .2 (95% CI: .1 to .3) comparing children with the most and least severe functional limitations; OR = 3.2 (95%CI: 2.1 to 4.8) comparing children with IQ < 70 and others; OR = 2.7 (95% CI: 1.5 to 4.6) comparing children in severe pain and others; OR = 2.7 (95% CI:1.6 to 4.6) comparing children with another disabled sibling or OR = 1.8 (95%CI: 1.2 to 2.8) no siblings and others; OR = 1.8 (95% CI: 1.1 to 2.8) comparing children resident in a town and others. Among parents who reported their child to have psychological problems, 95% said they had lasted over a year, 37% said they distressed their child and 42% said they burdened the family at least ‘quite a lot’.

Conclusions: A significant proportion of children with cerebral palsy have psychological symptoms or social impairment sufficiently severe to warrant referral to specialist services. Care must be taken in the assessment and management of children with cerebral palsy to ensure psychological problems are not overlooked and potentially preventable risk factors like pain are treated effectively. The validity of the SDQ for children with severe disability warrants further assessment.

SN 2008S: an electron-capture SN from a super-AGB progenitor?

We present comprehensive photometric and spectroscopic observations of the faint transient SN 2008S discovered in the nearby galaxy NGC 6946. SN 2008S exhibited slow photometric evolution and almost no spectral variability during the first nine months, implying a long photon diffusion time and a high-density circumstellar medium. Its bolometric luminosity (similar or equal to 10(41) erg s(-1) at peak) is low with respect to most core-collapse supernovae but is comparable to the faintest Type II-P events. Our quasi-bolometric light curve extends to 300 d and shows a tail phase decay rate consistent with that of Co-56. We propose that this is evidence for an explosion and formation of Ni-56 (0.0014 +/- 0.0003 M-circle dot). Spectra of SN 2008S show intense emission lines of H alpha, [Ca II] doublet and Ca II near-infrared (NIR) triplet, all without obvious P-Cygni absorption troughs. The large mid-infrared (MIR) flux detected shortly after explosion can be explained by a light echo from pre-existing dust. The late NIR flux excess is plausibly due to a combination of warm newly formed ejecta dust together with shock-heated dust in the circumstellar environment. We reassess the progenitor object detected previously in Spitzer archive images, supplementing this discussion with a model of the MIR spectral energy distribution. This supports the idea of a dusty, optically thick shell around SN 2008S with an inner radius of nearly 90 AU and outer radius of 450 AU, and an inferred heating source of 3000 K. The luminosity of the central star is L similar or equal to 10(4.6) L-circle dot. All the nearby progenitor dust was likely evaporated in the explosion leaving only the much older dust lying further out in the circumstellar environment. The combination of our long-term multiwavelength monitoring data and the evidence from the progenitor analysis leads us to support the scenario of a weak electron-capture supernova explosion in a super-asymptotic giant branch progenitor star (of initial mass 6-8 M-circle dot) embedded within a thick circumstellar gaseous envelope. We suggest that all of main properties of the electron-capture SN phenomenon are observed in SN 2008S and future observations may allow a definitive answer.

WASP-14b: 7.3 M-J transiting planet in an eccentric orbit

We report the discovery of a 7.3 M-J exoplanet WASP-14b, one of the most massive transiting exoplanets observed to date. The planet orbits the 10th-magnitude F5V star USNO-B1 11118-0262485 with a period of 2.243 752 d and orbital eccentricity e = 0.09. A simultaneous fit of the transit light curve and radial velocity measurements yields a planetary mass of 7.3 +/- 0.5 M-J and a radius of 1.28 +/- 0.08 R-J. This leads to a mean density of about 4.6 g cm(-3) making it the densest transiting exoplanets yet found at an orbital period less than 3 d. We estimate this system to be at a distance of 160 +/- 20 pc. Spectral analysis of the host star reveals a temperature of 6475 +/- 100 K, log g = 4.07 cm s(-2) and v sin i = 4.9 +/- 1.0 km s(-1), and also a high lithium abundance, log N(Li) = 2.84 +/- 0.05. The stellar density, effective temperature and rotation rate suggest an age for the system of about 0.5-1.0 Gyr.

Strong erbium luminescence in the near-infrared telecommunication window

A new type of near-infrared emitting rare-earth complex has been synthesised, consisting of three bis(perfluoroalkylsulfonyl)imide ligands and one 1,10-phenanthroline molecule. The chelate rings formed by the rare-earth ion and the bidentate ligands do not contain any carbon atoms and can hence be considered as 'inorganic' chelate rings. The absence of C-H stretching vibration modes in the first coordination sphere of the rare-earth ion and the presence of a light-harvesting moiety (1,10-phenanthroline) bound to the rare-earth ion result in a complex that can be efficiently excited and exhibits intense near-infrared luminescence. (C) 2004 Elsevier B.V. All rights reserved.

Ionic liquids as solvents for near-infrared emitting lanthanide complexes

It is shown that ionic liquids are promising solvents for near-infrared emitting lanthanide complexes, because ionic liquids are polar non-coordinating solvents that can solubilize lanthanide complexes. Neodymium(III) tosylate, bromide, triflate and sulfonylimide complexes were dissolved in 1-alkyl-3-methylimidazolium ionic liquids that contain the same anion as the neodymium(III) complexes. Near-infrared luminescence spectra of these neodymium(III) salts were measured by direct excitation of the neodymium(III) ion. The absorption spectra show detailed crystal-field fine structure and Judd-Ofelt parameters have been determined. Intense near-infrared luminescence was observed upon ligand excitation for neodymium(III) complexes with 1,10-phenanthroline or beta-diketonate ligands. (C) 2004 Elsevier B.V. All rights reserved.

Ionic Liquids Containing Boron Cluster Anions

The combination of different boron cluster anions and some of the cations typically found in the composition of ionic liquids has been possible by straightforward metathetic reactions, producing new low melting point salts; the imidazolium cations have been systematically studied, [C(n)mim](+) (when [C(n)mim](+) = 1-alkyl-3-methylimidazolium; n = 2, 4, 6, 8, 10, 12, 14, 16, or 18). Melting points increase in the anionic order [Co(C2B9H11)(2)](-) =-34 degrees C). The salts [C(n)mim](2)[X] ([X](2-) = [B10Cl10](2-) or [B12Cl12](2-), n = 16 or 18) show liquid crystal phases between the solid and liquid states. Tetraalkylphosphonium salts of [B10Cl10](2-) have also been prepared. Physical properties, such as thermal stability, density, or viscosity, have been measured for some selected samples. The presence of the perhalogenated dianion [B12Cl12](2-) in the composition of the imidazolium salts renders highly thermally stable compounds. For example, [C(2)mim](2)[B12Cl12] starts to decompose above 480 degrees C in a dynamic TGA analysis under a dinitrogen atmosphere. Crystal structures of [C(2)mim][Co(C2B9H11)(2)] and [C(2)mim](2)[B12Cl12] have been determined. H-1 NMR spectra of selected imidazolium-boron cluster anion salts have been recorded from solutions as a function of the concentration, showing trends related to the cation-anion interactions.

Electron attachment to 2-nitro-m-xylene

Electron attachment to nitroaromatic compound 2-nitro-m-xylene in gas phase has been performed utilizing a double focusing two sector mass spectrometer with high mass resolution (m/Delta m approximate to 2500). At low energy (below 20 eV), electron interactions with the neutral 2-nitro-m-xylene molecule reveal a very rich fragmentation pattern. A total of 60 fragment anions have been detected and the ion yield for all observed negative ions has been recorded as a function of the incident electron energy, among them a long lived (metastable) non-dissociated parent anion which is formed at energies near zero eV, and some ions observed at the mass numbers 26,42 and 121. Comparison of calculated isotopic patterns with measured ion yields for these fragment anions and their successors in the mass spectrum, allows the assignment of the chemical composition of these fragments as CN- (26 Da), CNO- (42 Da) and C8H9O- (121 Da). Electron attachment to 2-nitro-m-xylene leads to anion formation at four energy ranges. Between 0 eV and 2 eV only few product ions are formed. Between 4.6 eV and 6.1 eV all fragment anions are formed and for most of them the anion yield reaches its maximum value in this range. NO2- which is the most abundant product [M-H](-) and O- are the only fragments that exhibit a feature at 7.4eV, 8.1 eV and 7.9eV, respectively. About half of the fragment anions exhibit a broad, mostly low-intensity resonance between 9 eV and 10 eV. (C) 2009 Elsevier B.V. All rights reserved.

PHOTOMINERALIZATION OF 4-CHLOROPHENOL SENSITIZED BY TITANIUM-DIOXIDE - A STUDY OF THE INITIAL KINETICS OF CARBON-DIOXIDE PHOTOGENERATION

The kinetics of photomineralization of 4-chlorophenol (4-CP) sensitized by Degussa P25 TiO2 in O2-saturated solution is studied as a function of the following different experimental parameters: pH, [TiO2], percentage O2 [O2], [4-CP], T, I, lambda and [KNO3]. At pH 2 and T=30-degrees-C the initial relative rate of CO2 photogeneration R(CO2) conforms to a Langmuir-Hinshelwood-type kinetic scheme and the relationship between R(CO2) and the various experimental parameters may be summarized as follows: R(CO2) = gammaK(O2)[O2](I(a))(theta)K(4-CP]0/(1 + K(O2])(1 + K(4-CP)[4-CP]0) where gamma is a proportionality constant, K(O2) = 0.044 +/- 0.005[O2]-1, theta = 0.74 +/- 0.05 and K(4-CP) = (29 +/- 3) x 10(3) dm3 mol-1. The overall activation energy for this photosystem was determined as 16 +/- 2 kJ mol-1. This work forms part of an overall characterization study in which it is proposed that the 4-CP-TiO2-O2 photosystem is adopted as a standard test system for incorporation into all future semiconductor-sensitized photomineralization studies in order to facilitate comparisons between the results of the different studies.

Inactivation of V79 cells by low-energy protons, deuterons and helium-3 ions

Previous work by ourselves and by others has demonstrated that protons with a linear energy transfer (LET) about 30 V mu m(-1) are more effective at killing cells than doubly charged particles of the same LET. In this work we show that by using deuterons, which have about twice the range of protons with the same LET, it is possible to extend measurements of the RBE of singly charged particles to higher LET (up to 50 keV mu m(-1)). We report the design and use of a new arrangement for irradiating V79 mammalian cells. Cell survival. measurements have been made using protons in the energy range 1.0-3.7 MeV, deuterons in the energy range 0.9-3.4 MeV and He-3(2+) ions in the energy range 3.4-6.9 MeV;. This corresponds to volume-averaged LET (within the cell nucleus) between 10 and 28 keV mu m(-1) for protons, 18-50 keV mu m(-1) for deuterons, and 59-106 keV mu m(-1) for helium ions. Our results show no difference in the effectiveness of protons and deuterons matched for LET. However, for LET above about 30 keV mu m(-1) singly charged particles are more effective at inactivating cells than doubly-charged particles of the same LET and that this difference can be understood in terms of the radial dose distribution around the primary ion track.

Low flow water quality in rivers; septic tank systems and high-resolution phosphorus signals

Rural point sources of phosphorus (P), including septic tank systems, provide a small part of the overall phosphorus budget to surface waters in agricultural catchments but can have a disproportionate impact on the low flow P concentration of receiving rivers. This has particular importance as the discharges are approximately constant into receiving waters and these have restricted dilution capacity during ecologically sensitive summer periods. In this study, a number of identified high impact septic systems were replaced with modern sequential batch reactors in three rural catchments during a monitoring period of 4 years. Sub-hourly P monitoring was conducted using bankside-analysers. Results show that strategic replacement of defective septic tank systems with modern systems and polishing filters decreased the low flow P concentration of one catchment stream by 0.032 mg TP L- 1 (0.018 mg TRP L- 1) over the 4 years. However two of the catchment mitigation efforts were offset by continued new-builds that increased the density of septic systems from 3.4 km- 2 to 4.6 km- 2 and 13.8 km- 2 to 17.2 km- 2 and subsequently increased low flow P concentrations. Future considerations for septic system mitigation should include catchment carrying capacity as well as technology changes.