Dynamics of on-line gradient descent learning for multilayer neural networks

We consider the problem of on-line gradient descent learning for general two-layer neural networks. An analytic solution is presented and used to investigate the role of the learning rate in controlling the evolution and convergence of the learning process.

Learning with noise and regularizers in multilayer neural networks

We study the effect of two types of noise, data noise and model noise, in an on-line gradient-descent learning scenario for general two-layer student network with an arbitrary number of hidden units. Training examples are randomly drawn input vectors labeled by a two-layer teacher network with an arbitrary number of hidden units. Data is then corrupted by Gaussian noise affecting either the output or the model itself. We examine the effect of both types of noise on the evolution of order parameters and the generalization error in various phases of the learning process.

Learning with regularizers in multilayer neural networks

We study the effect of regularization in an on-line gradient-descent learning scenario for a general two-layer student network with an arbitrary number of hidden units. Training examples are randomly drawn input vectors labelled by a two-layer teacher network with an arbitrary number of hidden units which may be corrupted by Gaussian output noise. We examine the effect of weight decay regularization on the dynamical evolution of the order parameters and generalization error in various phases of the learning process, in both noiseless and noisy scenarios.

Effect of vesicle size on tissue localization and immunogenicity of liposomal DNA vaccines

The formulation of plasmid DNA (pDNA) in cationic liposomes is a promising strategy to improve the potency of DNA vaccines. In this respect, physicochemical parameters such as liposome size may be important for their efficacy. The aim of the current study was to investigate the effect of vesicle size on the in vivo performance of liposomal pDNA vaccines after subcutaneous vaccination in mice. The tissue distribution of cationic liposomes of two sizes, 500 nm (PDI 0.6) and 140 nm (PDI 0.15), composed of egg PC, DOPE and DOTAP, with encapsulated OVA-encoding pDNA, was studied by using dual radiolabeled pDNA-liposomes. Their potency to elicit cellular and humoral immune responses was investigated upon application in a homologous and heterologous vaccination schedule with 3 week intervals. It was shown that encapsulation of pDNA into cationic lipsomes resulted in deposition at the site of injection, and strongest retention was observed at large vesicle size. The vaccination studies demonstrated a more robust induction of OVA-specific, functional CD8+ T-cells and higher antibody levels upon vaccination with small monodisperse pDNA-liposomes, as compared to large heterodisperse liposomes or naked pDNA. The introduction of a PEG-coating on the small cationic liposomes resulted in enhanced lymphatic drainage, but immune responses were not improved when compared to non-PEGylated liposomes. In conclusion, it was shown that the physicochemical properties of the liposomes are of crucial importance for their performance as pDNA vaccine carrier, and cationic charge and small size are favorable properties for subcutaneous DNA vaccination.

The effect of rock surface aspect on growth, size structure and competition in the lichen Rhizocarpon geographicum

Rhizocarpon geographicum is a crustose lichen found frequently on rock surfaces of southern aspect and less frequently on rock surfaces of northern aspect in Gwynedd, North Wales. This study tested the hypothesis that the radial growth of R. geographicum thalli predicts aspect distribution. Thalli of all sizes, however, exhibited significantly greater radial growth over 18 months on northwest compared with southeast facing surfaces. The hypothesis that a more intense competitive environment on northwest facing surfaces may explain the aspect distribution of R. geographicum was then tested. The size frequency distributions of thalli revealed a higher proportion of thalli in the smallest size class and a more restricted thallus size range on the northwest facing surfaces. In addition, thallus mortality appeared to be greater on northwest facing surfaces. Significantly more associated lichen species were present on rock surfaces of northern aspect at sites where R. geographicum was present. The mean frequency of the associated lichen species, however, was significantly lower on surfaces of northern aspect where R. geographicum was present. In addition, two common foliose species at these sites were demonstrated experimentally to overgrow thalli of R. geographicum. It is concluded that the growth of R. geographicum over the study period did not predict aspect distribution and that differences in the competitive environments on northwest and southeast surfaces may be an important factor determining aspect distribution. © 2002 Elsevier Science B.V. All rights reserved.

Blurred edges look faint, and faint edges look sharp:The effect of a gradient threshold in a multi-scale edge coding model

A multi-scale model of edge coding based on normalized Gaussian derivative filters successfully predicts perceived scale (blur) for a wide variety of edge profiles [Georgeson, M. A., May, K. A., Freeman, T. C. A., & Hesse, G. S. (in press). From filters to features: Scale-space analysis of edge and blur coding in human vision. Journal of Vision]. Our model spatially differentiates the luminance profile, half-wave rectifies the 1st derivative, and then differentiates twice more, to give the 3rd derivative of all regions with a positive gradient. This process is implemented by a set of Gaussian derivative filters with a range of scales. Peaks in the inverted normalized 3rd derivative across space and scale indicate the positions and scales of the edges. The edge contrast can be estimated from the height of the peak. The model provides a veridical estimate of the scale and contrast of edges that have a Gaussian integral profile. Therefore, since scale and contrast are independent stimulus parameters, the model predicts that the perceived value of either of these parameters should be unaffected by changes in the other. This prediction was found to be incorrect: reducing the contrast of an edge made it look sharper, and increasing its scale led to a decrease in the perceived contrast. Our model can account for these effects when the simple half-wave rectifier after the 1st derivative is replaced by a smoothed threshold function described by two parameters. For each subject, one pair of parameters provided a satisfactory fit to the data from all the experiments presented here and in the accompanying paper [May, K. A. & Georgeson, M. A. (2007). Added luminance ramp alters perceived edge blur and contrast: A critical test for derivative-based models of edge coding. Vision Research, 47, 1721-1731]. Thus, when we allow for the visual system's insensitivity to very shallow luminance gradients, our multi-scale model can be extended to edge coding over a wide range of contrasts and blurs. © 2007 Elsevier Ltd. All rights reserved.

A CFD approach on the effect of particle size on char entrainment in bubbling fluidised bed reactors

The fluid – particle interaction inside a 41.7 mg s-1 fluidised bed reactor is modelled. Three char particles of sizes 500 µm, 250 µm, and 100 µm are injected into the fluidised bed and the momentum transport from the fluidising gas and fluidised sand is modelled. Due to the fluidising conditions and reactor design the char particles will either be entrained from the reactor or remain inside the bubbling bed. The particle size is the factor that differentiates the particle motion inside the reactor and their efficient entrainment out of it. A 3-Dimensional simulation has been performed with a completele revised momentum transport model for bubble three-phase flow according to the literature as an extension to the commercial finite volume code FLUENT 6.2.

The vesicle size of DDA:TDB liposomal adjuvants plays a role in the cell-mediated immune response but has no significant effect on antibody production

The use of cationic liposomes as experimental adjuvants for subunit peptide of protein vaccines is well documented. Recently the cationic liposome CAF01, composed of dimethyldioctadecylammonium (DDA) and trehalose dibehenate (TDB), has entered Phase I clinical trials for use in a tuberculosis (TB) vaccine. CAF01 liposomes are a heterogeneous population with a mean vesicle size of 500 nm; a strong retention of antigen at the injection site and a Th1-biassed immune response are noted. The purpose of this study was to investigate whether CAF01 liposomes of significantly different vesicle sizes exhibited altered pharmacokinetics in vivo and cellular uptake with activation in vitro. Furthermore, the immune response against the TB antigen Ag85B-ESAT-6 was followed when various sized CAF01 liposomes were used as vaccine adjuvants. The results showed no differences in vaccine (liposome or antigen) draining from the injection site, however, significant differences in the movement of liposomes to the popliteal lymph node were noted. Liposome uptake by THP-1 vitamin D3 stimulated macrophage-like cells did not show a liposome size-dependent pattern of uptake. Finally, whilst there were no significant differences in the IgG1/2 regardless of the liposome size used as a delivery vehicle for Ag85B-ESAT-6, vesicle size has a size dependent effect on cell proliferation and IL-10 production with larger liposomes (in excess of 2 µm) promoting the highest proliferation and lowest IL-10 responses, yet vesicles of ~500 nm promoting higher IFN-? cytokine production from splenocytes and higher IL-1ß at the site of injection.

The postantibiotic effect of the carbapenem antibiotics on gram-negative bacteria

Postantibiotic effect (PAE) describes the suppression of microbial growth occurring after a short exposure to an antimicrobial agent. PAE appears to be a property of the majority of antimicrobial agents and is demonstrated by a wide variety of microorganisms. At present, carbapenems and penems are the only members of the -lactam group of antimicrobial agents that exhibit a significant PAE on Gram-negative bacilli. A standardised method was developed to evaluate the in vitro PAE of three carbapenems; imipenem, meropenem and biapenem on Gram-negative bacteria under reproducible laboratory conditions that partially mimicked those occurring in vivo. The effects on carbapenem PAE of the method of antimicrobial removal, concentration, exposure duration, inoculum size, inoculum growth phase, multiple exposures and pooled human serum were determined. Additionally, the reproducibility, susceptibility prior to and after PAE determination and inter-strain variation of carbapenem PAE were evaluated. The method developed determined PAE by utilising viable counts and demonstrated carbapenem PAE to be reproducible, constant over successive exposures, dependent on genera, concentration, duration of exposure, inoculum size and growth phase. In addition, carbapenem PAE was not significantly effected either by agitation, the antimicrobial removal method or the viable count diluent. At present, the mechanism underlying PAE is undetermined. It is thought to be due to either the prolonged persistence of the antimicrobial at the cellular site of action or the true recovery period from non-lethal damage. Increasing the L-lysine concentration and salinity at recovery decreased and increased the carbapenem and imipenem PAE of Pseudomonas aeruginosa, respectively. In addition, no apparent change was observed in the production of virulence factors by P.aeruginosa in PAE phase. However, alterations in cell morphology were observed throughout PAE phase, and the reappearance of normal cell morphology corresponded to the duration of PAE determined by viable count. Thus, the recovery of the penicillin binding protein target enzymes appears to be the mechanism behind carbapenem PAE in P. aeruginosa.