A Note on the Generalization Performance of Kernel Classifiers with Margin

We present distribution independent bounds on the generalization misclassification performance of a family of kernel classifiers with margin. Support Vector Machine classifiers (SVM) stem out of this class of machines. The bounds are derived through computations of the $V_gamma$ dimension of a family of loss functions where the SVM one belongs to. Bounds that use functions of margin distributions (i.e. functions of the slack variables of SVM) are derived.

Sparse Correlation Kernel Analysis and Reconstruction

This paper presents a new paradigm for signal reconstruction and superresolution, Correlation Kernel Analysis (CKA), that is based on the selection of a sparse set of bases from a large dictionary of class- specific basis functions. The basis functions that we use are the correlation functions of the class of signals we are analyzing. To choose the appropriate features from this large dictionary, we use Support Vector Machine (SVM) regression and compare this to traditional Principal Component Analysis (PCA) for the tasks of signal reconstruction, superresolution, and compression. The testbed we use in this paper is a set of images of pedestrians. This paper also presents results of experiments in which we use a dictionary of multiscale basis functions and then use Basis Pursuit De-Noising to obtain a sparse, multiscale approximation of a signal. The results are analyzed and we conclude that 1) when used with a sparse representation technique, the correlation function is an effective kernel for image reconstruction and superresolution, 2) for image compression, PCA and SVM have different tradeoffs, depending on the particular metric that is used to evaluate the results, 3) in sparse representation techniques, L_1 is not a good proxy for the true measure of sparsity, L_0, and 4) the L_epsilon norm may be a better error metric for image reconstruction and compression than the L_2 norm, though the exact psychophysical metric should take into account high order structure in images.

On the V(subscript gamma) Dimension for Regression in Reproducing Kernel Hilbert Spaces

This paper presents a computation of the $V_gamma$ dimension for regression in bounded subspaces of Reproducing Kernel Hilbert Spaces (RKHS) for the Support Vector Machine (SVM) regression $epsilon$-insensitive loss function, and general $L_p$ loss functions. Finiteness of the RV_gamma$ dimension is shown, which also proves uniform convergence in probability for regression machines in RKHS subspaces that use the $L_epsilon$ or general $L_p$ loss functions. This paper presenta a novel proof of this result also for the case that a bias is added to the functions in the RKHS.

A comparison of the alr and ilr transformations for kernel density estimation of compositional data

In a seminal paper, Aitchison and Lauder (1985) introduced classical kernel density estimation techniques in the context of compositional data analysis. Indeed, they gave two options for the choice of the kernel to be used in the kernel estimator. One of these kernels is based on the use the alr transformation on the simplex SD jointly with the normal distribution on RD-1. However, these authors themselves recognized that this method has some deficiencies. A method for overcoming these dificulties based on recent developments for compositional data analysis and multivariate kernel estimation theory, combining the ilr transformation with the use of the normal density with a full bandwidth matrix, was recently proposed in Martín-Fernández, Chacón and Mateu- Figueras (2006). Here we present an extensive simulation study that compares both methods in practice, thus exploring the finite-sample behaviour of both estimators

Loss of nerve fiber layer in HIV positive patients and comparative patterns in a series of cases in Bucaramanga, Colombia

Objective: The objective of this study is to conduct a description of the features of optic neuropathy associated with Human Immunodeficiency Virus in relation to their possible incidence within our population, regarding that there is no data in our population in terms of frequency of this pathology (1,2). Methodology: Descriptive cross-sectional study of a clinical series of patients infected with human immunodeficiency virus, but AIDS, and the thickness of optic nerve´s layer of fibers studied with OCT technology (optical coherence tomography), patients were cited once captured. OCT was performed by the same observer, by taking 3 shots and picking the one with better reliability. Patients were given personally to the Ophthalmologic Foundation of Santander to conduct the review called OCT (optical coherence tomography). Results: In terms of viral load variable, we found a clear correlation in which validates the hypothesis that lower viral load means a thicker layer of fibers finding statistically significant differences for the 6 hours in right eye and 12 and 6 hours in left eye. Comparison between the known nomogram of fiber layer thickness for the population of Bucaramanga, Santander and thickness found in our sample, we note a clear decrease in the upper and lower quadrants, specifically in 7 hours and 11 hours, being more important in 7 hours, showing statistically significant differences. Conclusions: The pattern  of thinning of the nerve fiber layer in HIV positive patients without AIDS, and antiretroviral treatment type HAART, showed a statistically significant thinning targeted at 7 hours and 11 hours, being higher in first. Viral load figures have a direct relation with loss fiber layer, showing a statistically significant difference for the 6 and 12 hours.

A Monte Carlo-Based Fiber Tracking Algorithm using Diffusion Tensor MRI

Diffusion tensor magnetic resonance imaging, which measures directional information of water diffusion in the brain, has emerged as a powerful tool for human brain studies. In this paper, we introduce a new Monte Carlo-based fiber tracking approach to estimate brain connectivity. One of the main characteristics of this approach is that all parameters of the algorithm are automatically determined at each point using the entropy of the eigenvalues of the diffusion tensor. Experimental results show the good performance of the proposed approach

Optimisation of protoplast production in white lupin

The influence, was investigated, of abiotic parameters on the isolation of protoplasts from in vitro seedling cotyledons of white lupin. The protoplasts were found to be competent in withstanding a wide range of osmotic potentials of the enzyme medium, however, -2.25 MPa (0.5 M mannitol), resulted in the highest yield of protoplasts. The pH of the isolation medium also had a profound effect on protoplast production. Vacuum infiltration of the enzyme solution into the cotyledon tissue resulted in a progressive drop in the yield of protoplasts. The speed and duration of orbital agitation of the cotyledon tissue played a significant role in the release of protoplasts and a two step (stationary-gyratory) regime was found to be better than the gyratory-only system.

A breakthrough in lupin biotechnology: prolific protocolonisation in recalcitrant white lupin (Lupinus albus) triggered by bovine serum albumin

Six nutrient formulations were studied for their efficacy in inducing mitosis in white lupin seedling cotyledon protoplasts of which the formulations of Schafer-Menuhr & Sturmer (AS) and Kao (K8p) were found to be superior over the other four when supplemented with 6-benzylaminopurine and alpha-naphthaleneacetic acid (alpha-NAA). An unltrafiltration treatment of K8p increased mitotic frequency by 130% when compared with the untreated control. Medium enrichment with 0.2% bovine serum albumin (BSA) brought about a dramatic 1341% rise in protoplast division in comparison with BSA-free medium but only when the enrichment was carried out in Kao and Michayluk (KM8p) background containing 2, 4-dichlorophenoxyacetic acid, alpha-NAA and zeatin. A higher number of protocolonies (each proliferating from single protoplast following multiple divisions) were seen in 0.4% BSA. With this breakthrough in white lupin protoplast research, it is now possible to reproducibly obtain protocolonies that was hitherto not possible.

Aspects of isolation underpinning mitotic behaviour in lupin protoplasts

This study reports on the influence of critical isolation factors on the subsequent culture of protoplasts of Lupinus albus L. Protoplasts were isolated from in vitro seedling cotyledons of five early maturing accessions in which protoplast yields and division frequencies appeared to be correlated as a high protoplast yield corresponded with a high division frequency. The overall difference among the accessions for mitosis was non- significant, although the highest yield and division frequency were observed in accession LA132, with Alban giving a significantly lower level. Accession Lucrop produced the lowest number of protoplasts, all of which collapsed during culture. Of the enzyme types used for tissue maceration, Pectolyase Y23, was significantly inferior to Macerase in terms of giving way to mitosis. The extent of division in Macerase- isolated protoplast population was 266% higher than that in the Pectolyase Y23- isolated one. The physiological maturity level of the explant, expressed in terms of developmental age, was optimal when 14 - 18- day- old seedling cotyledons were used for protoplast production and culture, rather than more mature ones, despite higher protoplast yields in the latter. On K8p medium, the protoplast division frequency was 129% greater when 18- day- old seedling cotyledons were used, than that with any other treatment. This work on protoplast culture of the potentially important lupin species, which is a pulse rich in dietary protein, oil and fibre, allows a further understanding of the biology, with an aim to advance lupin biotechnology.

Optimization of protoplast production in white lupin

The influence, was investigated, of abiotic parameters on the isolation of protoplasts from in vitro seedling cotyledons of white lupin. The protoplasts were found to be competent in withstanding a wide range of osmotic potentials of the enzyme medium, however, −2.25 MPa (0.5 M mannitol), resulted in the highest yield of protoplasts. The pH of the isolation medium also had a profound effect on protoplast production. Vacuum infiltration of the enzyme solution into the cotyledon tissue resulted in a progressive drop in the yield of protoplasts. The speed and duration of orbital agitation of the cotyledon tissue played a significant role in the release of protoplasts and a two step (stationary-gyratory) regime was found to be better than the gyratory-only system.

Structure and biogenesis of the capsular F1 antigen from Yersinia pestis: Preserved folding energy drives fiber formation

Most gram-negative pathogens express fibrous adhesive virulence organelles that mediate targeting to the sites of infection. The F1 capsular antigen from the plague pathogen Yersinia pestis consists of linear fibers of a single subunit (Caf1) and serves as a prototype for nonpilus organelles assembled via the chaperone/usher pathway. Genetic data together with high-resolution X-ray structures corresponding to snapshots of the assembly process reveal the structural basis of fiber formation. Comparison of chaperone bound Caf1 subunit with the subunit in the fiber reveals a novel type of conformational change involving the entire hydrophobic core of the protein. The observed conformational change suggests that the chaperone traps a high-energy folding intermediate of Caf1. A model is proposed in which release of the subunit allows folding to be completed, driving fiber formation.