The use of confidence measures to enhance combination strategies in multi-network neuro-fuzzy systems

It is well known that substantial improvements can be obtained in difficult pattern recognition problems by combining or integrating the outputs of multiple neural classifiers. This paper analyses the performance of some combination schemes applied to a multi-hybrid neural system which is composed of neural and fuzzy neural networks. Essentially, the combination methods employ different ways to extract valuable information from the output of the experts through the use of confidence (weights) measures of the ensemble members to each class. An empirical evaluation in a handwritten numeral recognition task is used to investigate the performance of the presented methods in comparison with some existing combination methods.     

Simulation of statistical variability in nano-CMOS transistors using drift-diffusion, Monte Carlo and non-equilibrium Green’s function techniques

In this paper, we present models and tools developed and used by the Device Modelling Group at the University of Glasgow to study statistical variability introduced by the discreteness of charge and matter in contemporary and future Nano-CMOS transistors. The models and tools, based on Drift-Diffusion (DD), Monte Carlo (MC) and Non-Equilibrium Green’s Function (NEGF) techniques, are encapsulated in the Glasgow 3D statistical ‘atomistic’ device simulator. The simulator can handle most of the known sources of statistical variability including Random Discrete Dopants (RDD), Line Edge Roughness (LER), Thickness Fluctuations in the Oxide (OTF) and Body (BTF), granularity of the Poly-Silicon (PSG), Metal Gate (MGG) and High-κ (HKG), and oxide trapped charges (OTC). The results of the statistical simulations are verified with respect to measurements carried out on fabricated devices. Predictions about the magnitude of the statistical variability in future generations of nano-CMOS devices are also presented.     

Rhodium Containing Magnetic Nanoparticles: Effective Catalysts for Hydrogenation and the 1,4-Addition of Boronic Acids

New efficient rhodium catalysts supported on superparamagnetic iron oxide nanoparticles (NPs) have been prepared using a novel method involving sulfonated triphenylphosphine ligands. They successfully promote the hydrogenation of olefins as well as the addition of arylboronic acids to dimethyl itaconate (ItMe₂) in water for up to 10 recycles. The catalysts were stable towards leaching of the metal complexes and were readily recovered by applying an external magnetic field.     

Activation of a phenolic resin-derived carbon in air using microwave thermogravimetry

A phenolic resin-derived carbon has been activated, in air, using a microwave thermogravimetric analyser. Controlled Rate processing, where the rate of mass loss is maintained at a predetermined level throughout the activation process by continuous alteration of the microwave power, was used to prevent thermal runaway. Using this approach, carbons, with micropore volumes up to 0.44 cm³ g^-1, were produced in less than 170 minutes. It was also found that the thermal history of the pre-cursor carbon greatly affected the efficiency by which it was heated by microwave energy.     

Towards the design of a scalable and commercially viable technique for plasmid purification using a methacrylate monolithic stationary phase

A monolithic stationary phase was prepared via free radical co‐polymerization of ethylene glycol dimethacrylate (EDMA) and glycidyl methacrylate (GMA) with pore diameter tailored specifically for plasmid binding, retention and elution. The polymer was functionalized with 2‐chloro‐N,N‐diethylethylamine hydrochloride (DEAE‐Cl) for anion‐exchange purification of plasmid DNA (pDNA) from clarified lysate obtained from E. coli DH5α‐pUC19 culture in a ribonuclease/protease‐free environment. Characterization of the monolithic resin showed a porous material, with 68% of the pores existing in the matrix having diameters above 300 nm. The final product isolated from a single‐stage 5 min anion‐exchange purification was a pure and homogeneous supercoiled (SC) pDNA with no gDNA, RNA and protein contamination as confirmed by ethidium bromide agarose gel electrophoresis (EtBr‐AGE), enzyme restriction analysis and sodium dodecyl sulfate–polyacrylamide gel electrophoresis. This non‐toxic technique is cGMP compatible and highly scalable for production of pDNA on a commercial level. Copyright © 2007 Society of Chemical Industry     

Effect of breed type,housing and feeding system on performance of growing pigs managed under organic conditions

BACKGROUND: There is a need for information on the performance and carcass quality of pigs under different organic management systems. This study compared Duroc‐sired progeny from three maternal breed types when kept either at pasture or in housing with an outdoor run and offered ad libitum concentrate either alone or with fodder beet or grass/clover silage as additional forage. RESULTS: Liveweight gain, feed intake and the proportion of forages consumed did not differ between genotypes. Carcass fatness of progeny was lowest for a ‘modern’ genotype (Camborough 12) and highest for a ‘traditional’ purebred genotype (Saddleback), with a ‘crossbred traditional’ genotype (Saddleback × Duroc) being intermediate (11.4, 14.3 and 13.4 mm P₂ respectively, standard error of mean (SEM) 0.27, P < 0.001). With a cereal‐based concentrate available ad libitum, intake of forages was low (<2% of dry matter intake). Although growth rate did not differ between housing systems, daily feed intake was greater at pasture (2.47 vs 2.22 kg meal equivalent, SEM 0.05, P < 0.001), giving poorer feed efficiency (P < 0.01). Pastured animals consumed less additional forage and had a higher killing‐out % but similar carcass fatness. CONCLUSION: For organic pig production to be financially sustainable, disadvantages arising from the genotype and/or rearing system chosen need to be offset by a market premium for the pigs produced. Copyright © 2007 Society of Chemical Industry     

Binding properties of peptidic affinity ligands for plasmid DNA capture and detection

Peptides constructed from α‐helical subunits of the Lac repressor protein (LacI) were designed then tailored to achieve particular binding kinetics and dissociation constants for plasmid DNA purification and detection. Surface plasmon resonance was employed for quantification and characterization of the binding of double stranded Escherichia coli plasmid DNA (pUC19) via the lac operon (lacO) to “biomimics” of the DNA binding domain of LacI. Equilibrium dissociation constants (K_D), association (k_a), and dissociation rates (k_d) for the interaction between a suite of peptide sequences and pUC19 were determined. K_D values measured for the binding of pUC19 to the 47mer, 27mer, 16mer, and 14mer peptides were 8.8 ± 1.3 × 10^?10 M, 7.2 ± 0.6 × 10^?10 M, 4.5 ± 0.5 × 10^?8 M, and 6.2 ± 0.9 × 10^?6 M, respectively. These findings show that affinity peptides, composed of subunits from a naturally occurring operon–repressor interaction, can be designed to achieve binding characteristics suitable for affinity chromatography and biosensor devices. © 2008 American Institute of Chemical Engineers AIChE J, 2009     

Oxidative stability of Echium plantagineum seed oil bodies

Echium plantagineum seed contains a highly polyunsaturated oil (approximately 14% linoleic acid, 10% γ‐linolenic acid, 33% α‐linolenic acid and 14% stearidonic acid); almost half of the fatty acids are omega‐3 fatty acids, so there is an interest in the possible health benefits of this oil, which, once extracted, is prone to oxidation. For the first time in reported literature, oil bodies (OBs), the organelles that store the oil in mature seed, were recovered from E. plantagineum seeds. The oxidative stability of these organelles ex vivo, dispersed in an aqueous continuous phase, was tested against processed E. plantagineum oil emulsions stabilised with either SDS or Tween 20. For both primary and secondary oxidation products the OBs were the most stable form of dispersed oil, and the dispersed systems were all more stable than bulk E. plantagineum oil after incubating at 40°C for 7 days. The possible reasons for the enhanced chemical stability of E. plantagineum OBs are explored in this paper. Practical applications: OBs, the natural store of oil in oilseeds, can be recovered from seeds intact and are relatively stable to oxidation ex vivo. Echium seed OBs, enriched in physiologically active omega‐3 fatty acids, therefore offer an attractive alternative to traditional oil extraction methods and overcome the need to encapsulate the omega‐3 rich oil.     

Use of density gradient quantum corrections in the simulation of statistical variability in MOSFETs

With the scaling of field-effect transistors to the nanometre scale, it is well recognised that TCAD simulations of such devices need to account for quantum mechanical confinement effects. The most widely used method to incorporate quantum effects within classical and semi-classical simulators is via density gradient quantum corrections. Here we present our methodologies for including the density gradient method within our Drift-Diffusion and Monte Carlo simulators and highlight some of the additional benefits that this provides when dealing with the charge associated with random discrete dopants.