Pattern classification with mixtures of weighted least-squares support vector machine experts

Support Vector Machine (SVM) classifiers are high-performance classification models devised to comply with the structural risk minimization principle and to properly exploit the kernel artifice of nonlinearly mapping input data into high-dimensional feature spaces toward the automatic construction of better discriminating linear decision boundaries. Among several SVM variants, Least-Squares SVMs (LS-SVMs) have gained increased attention recently due mainly to their computationally attractive properties coming as the direct result of applying a modified formulation that makes use of a sum-squared-error cost function jointly with equality, instead of inequality, constraints. In this work, we present a flexible hybrid approach aimed at augmenting the proficiency of LS-SVM classifiers with regard to accuracy/generalization as well as to hyperparameter calibration issues. Such approach, named as Mixtures of Weighted Least-Squares Support Vector Machine Experts, centers around the fusion of the weighted variant of LS-SVMs with Mixtures of Experts models. After the formal characterization of the novel learning framework, simulation results obtained with respect to both binary and multiclass pattern classification problems are reported, ratifying the suitability of the novel hybrid approach in improving the performance issues considered.     

Free radical polymerization of ethyl methacrylate and ethyl α-hydroxy methacrylate: A computational approach to the propagation kinetics

The propagation kinetics of ethyl methacrylate (EMA) and ethyl α-hydroxy methacrylate (EHMA) has been subject to a computational study to understand their free radical polymerization (FRP) behavior in bulk and in solution using Density Functional Theory (DFT). The propagation of EHMA is studied in ethanol and toluene to assess the effect of hydrogen-bonding solvents on FRP of monomers with α-hydroxy functionality. Although EMA and EHMA resemble each other in structure, EHMA propagates faster in bulk due to the presence of intermolecular hydrogen-bonds, which tend to facilitate the approach of the propagating species. This falls in contrast with the experimentally observed lower propagation rates of EHMA in ethanol compared to toluene. Calculations show that the 2.28 rate acceleration in toluene is governed by the ratio of the pre-exponential factors, which reflect the entropies of activation, in both media. The polar protic solvent ethanol has a disruptive effect via hydrogen-bonding on the 6-membered ring shape of EHMA monomers thus decreasing the entropy of activation of the reaction. In the case of toluene, there are no special interactions with the hydrophobic solvent, the entropy of activation is higher than in ethanol.     

Sugarcane bagasse cellulose/HDPE composites obtained by extrusion

Natural fibers used in this study were both pre-treated and modified residues from sugarcane bagasse. Polymer of high density polyethylene (HDPE) was employed as matrix in to composites, which were produced by mixing high density polyethylene with cellulose (10%) and Cell/ZrO₂·nH₂O (10%), using an extruder and hydraulic press. Tensile tests showed that the Cell/ZrO₂·nH₂O (10%)/HDPE composites present better tensile strength than cellulose (10%)/HDPE composites. Cellulose agglomerations were responsible for poor adhesion between fiber and matrix in cellulose (10%)/HDPE composites. HDPE/natural fibers composites showed also lower tensile strength in comparison to the polymer. The increase in Young’s modulus is associated to fibers reinforcement. SEM analysis showed that the cellulose fibers insertion in the matrix caused an increase of defects, which were reduced when modified cellulose fibers were used.     

Hybridizing mixtures of experts with support vector machines: Investigation into nonlinear dynamic systems identification

Mixture of experts (ME) models comprise a family of modular neural network architectures aiming at distilling complex problems into simple subtasks. This is done by deploying a separate gating module for softly dividing the input space into overlapping regions to be each assigned to one or more expert networks. Conversely, support vector machines (SVMs) refer to kernel-based methods, neural-network-alike models that constitute an approximate implementation of the structural risk minimization principle. Such learning machines follow the simple, but powerful idea of nonlinearly mapping input data into high-dimensional feature spaces wherein a linear decision surface discriminating different regions is properly designed. In this work, we formally characterize and empirically evaluate a novel approach, named as Mixture of Support Vector Machine Experts (MSVME), whose main purpose is to combine the complementary properties of both SVM and ME models. In the formal characterization, an algorithm based on a maximum likelihood criterion is considered for the MSVME training, and we demonstrate that it is possible to train each expert based on an SVM perspective. Regarding the empirical evaluation, simulation results involving nonlinear dynamic system identification problems are reported, contrasting the performance shown by the MSVME approach with that exhibited by conventional SVM and ME models.     

Approach-withdrawal and cerebral asymmetry: Emotional expression and brain physiology: I.

In this experiment, we combined the measurement of observable facial behavior with simultaneous measures of brain electrical activity to assess patterns of hemispheric activation in different regions during the experience of happiness and disgust. Disgust was found to be associated with right-sided activation in the frontal and anterior temporal regions compared with the happy condition. Happiness was accompanied by left-sided activation in the anterior temporal region compared with disgust. No differences in asymmetry were found between emotions in the central and parietal regions. When data aggregated across positive films were compared to aggregate negative film data, no reliable differences in brain activity were found. These findings illustrate the utility of using facial behavior to verify the presence of emotion, are consistent with the notion of emotion-specific physiological patterning, and underscore the importance of anterior cerebral asymmetries for emotions associated with approach and withdrawal. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Influence of oxygen vacancies on the structural,dielectric, and magnetic properties of (Mn,Co) co-doped ZnO nanostructures

We analysed the variation and effect of oxygen vacancies on the structural, dielectric and magnetic properties in case of Mn (4%) and Co (1, 2 and 4%) co-doped ZnO nanoparticles (NPs), synthesized by chemical precipitation route and annealed at 750 °C for 2 h. From the XRD, the calculated average crystallite size increased from15.30?±?0.73 nm to 16.71?±?012 nm, when Co content is increased from 1 to 4%. Enhancement of dopants (Mn, Co) introduced more and more oxygen vacancies to ZnO lattice confirmed from EDX and XPS. The high-temperature annealing leads to reduction of the dielectric properties due to enhancement in grain growth (large grain volume and lesser number of grain boundaries) with the incorporation of Co and Mn ions into the ZnO lattice. The electrical conductivity of the Mn doped and (Mn, Co) co-doped ZnO samples were enhanced due to increase in the volume of conducting grains and charge density (liberation of trapped charge carriers in oxygen vacancies and free charge carriers at higher frequencies). The Mn-doped and (Mn, Co) co-doped ZnO NPs show ferromagnetic (FM) behaviour. The saturation and remnant magnetizations (M_s and M_r) elevates from (0.235 to 1.489)?×?10^?2 and (0.12 to 0.27)?×?10^?2 emu/g while Coercivity (H_c) reduced from 97 to 36 Oe with enhancement in the concentration of dopants in ZnO matrix. Oxygen vacancies were found to be the main reason for room-temperature ferromagnetism (RTFM) in the doped and co-doped ZnO NPs. The results show that the enhanced dielectric and magnetic properties of Mn doped and (Mn, Co) co-doped ZnO is strongly correlated with the concentration of oxygen vacancies. The observed enhanced RTFM, dielectric properties and electrical conductivity makes TM doped ZnO nanoparticles suitable for spintronics, microelectronics and optoelectronics based applications.     

Visual evoked potential measures of interhemispheric transfer time in humans.

Four experiments testing right-handed adult males examined interhemispheric transfer time (IHTT) estimation with visual evoked potentials (EPs) elicited in response to hemiretinal presentations of checkerboard-flash stimuli. Experiment 1 was a study of the relation between reaction time (RT) and EP measures of IHTT. EP measures provided more valid estimates than RT measures because more subjects showed IHTT in the direction of anatomical prediction. Experiment 2 showed that EPs derived from lateral occipital sites provided more valid and longer IHTT compared with EPs from medial occipital sites. Experiment 3 showed no difference between random versus blocked hemiretinal stimuli. Experiment 4 showed that IHTT derived with a linked-ears reference provided more valid estimates than IHTT derived with a mid-frontal reference and that small changes in stimulus eccentricity did not influence IHTT. The findings of these experiments indicate that noninvasive estimates of visual IHTT can be obtained in humans. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Development and characterization of a microthermoelectric generator with plated copper/constantan thermocouples

This work reports the development and the characterization of a microthermoelectric generator (μTEG) based on planar technology using electrochemically deposited constantan and copper thermocouples on a micro machined silicon substrate with a SiO₂/Si₃N₄/SiO₂ thermally insulating membrane to create a thermal gradient. The μTEG has been designed and optimized by finite element simulation in order to exploit the different thermal conductivity of silicon and membrane in order to obtain the maximum temperature difference on the planar surface between the hot and cold junctions of the thermocouples. The temperature difference was dependent on the nitrogen (N₂) flow velocity applied to the upper part of the device. The fabricated thermoelectric generator presented maximum output voltage and power of 118 mV/cm² and of 1.1 μW/cm², respectively, for a device with 180 thermocouples, 3 kΩ of internal resistance, and under a N₂ flow velocity of 6 m/s. The maximum efficiency (performance) was 2 × 10^?3 μW/cm² K².     

Stability of fruit juice drinks in aseptic packages

Cashew and passion fruit juice drinks were aseptically filled in cartons and in three types of plastic packages (with oxygen transmission rate‐OTR of 1,40; 2,96 and 13,74 cm³ (STP)/m²/day at 25 °C, 75% RH and 1 atm), with nitrogen gas flushing. No adverse effect on product sterility was found, neither were any significant changes in pH and acidity detected during storage at 25 ± 2°C. A decrease of the oxygen concentration in the headspace of the cashew juice drink plastic packs was observed, which appears to be caused by consumption of that gas in oxidation reactions. The shelf‐life of the cashew juice drink was limited to 45–70 days, after which period sensory changes attained objectionable levels (development of dark colours and deterioration of the typical fruit flavour). Passion fruit juice drink proved to be less sensitive to oxygen than cashew, since an increase of the oxygen concentration was detected in the headspace of the three plastic packs. The shelf‐life of passion fruit juice drink was limited to 70–90 days, due to deterioration of the typical passion fruit flavour and colour. Copyright © 2001 John Wiley & Sons, Ltd.     

The effective role of time in synthesising InN by chemical method at low temperature

This study involves the synthesise of indium nitride (InN) nanoparticles at low temperature using a chemical method. Three samples were synthesised under different times to produce InN nanoparticle of high quality crystallinity. Results showed that the time of synthesise plays an important role for N enhancement in InN nanoparticle structure. The average diameters of cubic phase of InN nanoparticle were 16.5 nm. These properties support the use of InN as a potential material for the manufacture of highly efficient low cost solar cells.