Support Vector Machines: Training and Applications

The Support Vector Machine (SVM) is a new and very promising classification technique developed by Vapnik and his group at AT&T Bell Labs. This new learning algorithm can be seen as an alternative training technique for Polynomial, Radial Basis Function and Multi-Layer Perceptron classifiers. An interesting property of this approach is that it is an approximate implementation of the Structural Risk Minimization (SRM) induction principle. The derivation of Support Vector Machines, its relationship with SRM, and its geometrical insight, are discussed in this paper. Training a SVM is equivalent to solve a quadratic programming problem with linear and box constraints in a number of variables equal to the number of data points. When the number of data points exceeds few thousands the problem is very challenging, because the quadratic form is completely dense, so the memory needed to store the problem grows with the square of the number of data points. Therefore, training problems arising in some real applications with large data sets are impossible to load into memory, and cannot be solved using standard non-linear constrained optimization algorithms. We present a decomposition algorithm that can be used to train SVM's over large data sets. The main idea behind the decomposition is the iterative solution of sub-problems and the evaluation of, and also establish the stopping criteria for the algorithm. We present previous approaches, as well as results and important details of our implementation of the algorithm using a second-order variant of the Reduced Gradient Method as the solver of the sub-problems. As an application of SVM's, we present preliminary results we obtained applying SVM to the problem of detecting frontal human faces in real images.

Aitchison Geometry for Probability and Likelihood as a new approach to mathematical statistics

The Aitchison vector space structure for the simplex is generalized to a Hilbert space structure A2(P) for distributions and likelihoods on arbitrary spaces. Central notations of statistics, such as Information or Likelihood, can be identified in the algebraical structure of A2(P) and their corresponding notions in compositional data analysis, such as Aitchison distance or centered log ratio transform. In this way very elaborated aspects of mathematical statistics can be understood easily in the light of a simple vector space structure and of compositional data analysis. E.g. combination of statistical information such as Bayesian updating, combination of likelihood and robust M-estimation functions are simple additions/ perturbations in A2(Pprior). Weighting observations corresponds to a weighted addition of the corresponding evidence. Likelihood based statistics for general exponential families turns out to have a particularly easy interpretation in terms of A2(P). Regular exponential families form finite dimensional linear subspaces of A2(P) and they correspond to finite dimensional subspaces formed by their posterior in the dual information space A2(Pprior). The Aitchison norm can identified with mean Fisher information. The closing constant itself is identified with a generalization of the cummulant function and shown to be Kullback Leiblers directed information. Fisher information is the local geometry of the manifold induced by the A2(P) derivative of the Kullback Leibler information and the space A2(P) can therefore be seen as the tangential geometry of statistical inference at the distribution P. The discussion of A2(P) valued random variables, such as estimation functions or likelihoods, give a further interpretation of Fisher information as the expected squared norm of evidence and a scale free understanding of unbiased reasoning

The four major N- and C-terminal splice variants of the excitatory amino acid transporter GLT-1 form cell surface homomeric and heteromeric assemblies

The L-glutamate transporter GLT-1 is an abundant CNS membrane protein of the excitatory amino acid transporter (EAAT) family which controls extracellular L-glutamate levels and is important in limiting excitotoxic neuronal death. Using RT-PCR, we have determined that four mRNAs encoding GLT-1 exist in mouse brain, with the potential to encode four GLT-1 isoforms that differ in their N- and C-termini. We expressed all four isoforms (termed MAST-KREK, MPK-KREK, MAST-DIETCI and MPK-DIETCI according to amino acid sequence) in a range of cell lines and primary astrocytes and show that each isoform can reach the cell surface. In transfected HEK-293 or COS-7 cells, all four isoforms support high-affinity sodium-dependent L-glutamate uptake with identical pharmacological and kinetic properties. Inserting a viral epitope (V5, HA or FLAG) into the second extracellular domain of each isoform allowed co-immunoprecipitation and tr-FRET studies using transfected HEK-293 cells. Here we show for the first time that each of the four isoforms are able to combine to form homomeric and heteromeric assemblies, each of which are expressed at the cell surface of primary astrocytes. After activation of protein kinase C by phorbol ester, V5-tagged GLT-1 is rapidly removed from the cell surface of HEK-293 cells and degraded. This study provides direct biochemical evidence for oligomeric assembly of GLT-1 and reports the development of novel tools to provide insight into the trafficking of GLT-1.

Predicting feed quality - chemical analysis and in vitro evaluation

As the ideal method of assessing the nutritive value of a feedstuff, namely offering it to the appropriate class of animal and recording the production response obtained, is neither practical nor cost effective a range of feed evaluation techniques have been developed. Each of these balances some degree of compromise with the practical situation against data generation. However, due to the impact of animal-feed interactions over and above that of feed composition, the target animal remains the ultimate arbitrator of nutritional value. In this review current in vitro feed evaluation techniques are examined according to the degree of animal-feed interaction. Chemical analysis provides absolute values and therefore differs from the majority of in vitro methods that simply rank feeds. However, with no host animal involvement, estimates of nutritional value are inferred by statistical association. In addition given the costs involved, the practical value of many analyses conducted should be reviewed. The in sacco technique has made a substantial contribution to both understanding rumen microbial degradative processes and the rapid evaluation of feeds, especially in developing countries. However, the numerous shortfalls of the technique, common to many in vitro methods, the desire to eliminate the use of surgically modified animals for routine feed evaluation, paralleled with improvements in in vitro techniques, will see this technique increasingly replaced. The majority of in vitro systems use substrate disappearance to assess degradation, however, this provides no information regarding the quantity of derived end-products available to the host animal. As measurement of volatile fatty acids or microbial biomass production greatly increases analytical costs, fermentation gas release, a simple and non-destructive measurement, has been used as an alternative. However, as gas release alone is of little use, gas-based systems, where both degradation and fermentation gas release are measured simultaneously, are attracting considerable interest. Alternative microbial inocula are being considered, as is the potential of using multi-enzyme systems to examine degradation dynamics. It is concluded that while chemical analysis will continue to form an indispensable part of feed evaluation, enhanced use will be made of increasingly complex in vitro systems. It is vital, however, the function and limitations of each methodology are fully understood and that the temptation to over-interpret the data is avoided so as to draw the appropriate conclusions. With careful selection and correct application in vitro systems offer powerful research tools with which to evaluate feedstuffs. (C) 2003 Elsevier B.V. All rights reserved.

Structure and function analysis of the poliovirus cis-acting replication element (CRE)

The poliovirus cis-acting replication element (CRE) templates the uridylylation of VPg, the protein primer for genome replication. The CRE is a highly conserved structural RNA element in the enteroviruses and located within the polyprotein-coding region of the genome. We have determined the native structure of the CRE, defined the regions of the structure critical for activity, and investigated the influence of genomic location on function. Our results demonstrate that a 14-nucleotide unpaired terminal loop, presented on a suitably stable stem, is all that is required for function. These conclusions complement the recent analysis of the 14-nucleotide terminal loop in the CRE of human rhinovirus type 14. The CRE can be translocated to the 5' noncoding region of the genome, at least 3.7-kb distant from the native location, without adversely influencing activity, and CRE duplications do not adversely influence replication. We do not have evidence for a specific interaction between the CRE and the RNA-binding 3CD(pro) complex, an essential component of the uridylylation reaction, and the mechanism by which the CRE is coordinated and orientated during the reaction remains unclear. These studies provide a detailed overview of the structural determinants required for CRE function, and will facilitate a better understanding of the requirements for picornavirus replication.

Modulation of the fecal microflora profile and immune function by a novel trans-galactooligosaccharide mixture (B-GOS) in healthy elderly volunteers

Background: Aging is associated with reduced numbers of beneficial colonic bifidobacteria and impaired immunity. Galactooligosaccharides (GOSs) stimulate the growth of bifidobacteria in younger adults, but little is known about their effects in the elderly and their immunomodulatory capacity. Objective: We assessed the effect of a prebiotic GOS mixture (B-GOS) on immune function and fecal microflora composition in healthy elderly subjects. Design: In a double-blind, placebo-controlled, crossover study, 44 elderly subjects were randomly assigned to receive either a placebo or the B-GOS treatment (5.5 g/d). Subjects consumed the treatments for 10 wk, and then went through a 4-wk washout period, before switching to the other treatment for the final 10 wk. Blood and fecal samples were collected at the beginning, middle (5 wk), and end of the test period. Predominant bacterial groups were quantified, and phagocytosis, natural killer (NK) cell activity, cytokine production, plasma cholesterol, and HDL cholesterol were measured. Results: B-GOS significantly increased the numbers of beneficial bacteria, especially bifidobacteria, at the expense of less beneficial groups compared with the baseline and placebo. Significant increases in phagocytosis, NK cell activity, and the production of antiinflammatory cytokine interleukin-10 (IL-10) and significant reduction in the production of proinflammatory cytokines (IL-6, IL-1 beta , and tumor necrosis factor-alpha) were also observed. B-GOS exerted no effects on total cholesterol or HDL-cholesterol production, however. Conclusions: B-GOS administration to healthy elderly persons resulted in positive effects on both the microflora composition and the immune response. Therefore, B-GOS may be a useful dietary candidate for the enhancement of gastrointestinal health and immune function in elderly persons. Am J Clin Nutr 2008; 88: 1438-46.

Green tea (Camellia sinensis) catechins and vascular function

The health benefits of green tea (Camellia sinensis) catechins are becoming increasingly recognised. Amongst the proposed benefits are the maintenance of endothelial function and vascular homeostasis and an associated reduction in atherogenesis and CVD risk. The mounting evidence for the influential effect of green tea catechins on vascular function from epidemiological, human intervention and animal studies is subject to review together with exploration of the potential mechanistic pathways involved. Epigallocatechin-3-gallate, one of the most abundant and widely studied catechin found in green tea, will be prominent in the present review. Since there is a substantial inconsistency in the published data with regards to the impact of green tea catechins on vascular function, evaluation and interpretation of the inter- and intra-study variability is included. In conclusion, a positive effect of green tea catechins on vascular function is becoming apparent. Further studies in animal and cell models using physiological concentrations of catechins and their metabolites are warranted in order to gain some insight into the physiology and molecular basis of the observed beneficial effects.

Isoflavones and endothelial function

Dietary isoflavones are thought to be cardioprotective due to their structural similarity to oestrogen. Oestrogen is believed to have beneficial effects on endothelial function and may be one of the mechanisms by which premenopausal women are protected against CVD. Decreased NO production and endothelial NO synthase activity, and increased endothelin-1 concentrations, impaired lipoprotein metabolism and increased circulating inflammatory factors result from oestrogen deficiency. Oestrogen acts by binding to oestrogen receptors alpha and beta. Isoflavones have been shown to bind with greater affinity to the latter. Oestrogen replacement therapy is no longer thought to be a safe treatment for prevention of CVD; isoflavones are a possible alternative. Limited evidence from human intervention studies suggests that isoflavones may improve endothelial function, but the available data are not conclusive. Animal studies provide stronger support for a role of isoflavones in the vasculature, with increased vasodilation and endothelial NO synthase activity demonstrated. Cellular mechanisms underlying the effects of isoflavones on endothelial cell function are not yet clear. Possible oestrogen receptor-mediated pathways include modulation of gene transcription, and also non-genomic oestrogen receptor-mediated signalling pathways. Putative non-oestrogenic pathways include inhibition of reactive oxygen species production and up regulation of the protein kinase A pathway (increasing NO bioavailability). Further research is needed to unravel effects of isoflavones on intracellular regulation of the endothelial function. Moreover, there is an urgent need for adequately powered, robustly designed human intervention studies in order to clarify the present equivocal findings.

Performance of GPRS coding scheme detection under severe multipath and co-channel interference as a function of soft-bit width

The General Packet Radio Service (GPRS) has been developed for the mobile radio environment to allow the migration from the traditional circuit switched connection to a more efficient packet based communication link particularly for data transfer. GPRS requires the addition of not only the GPRS software protocol stack, but also more baseband functionality for the mobile as new coding schemes have be en defined, uplink status flag detection, multislot operation and dynamic coding scheme detect. This paper concentrates on evaluating the performance of the GPRS coding scheme detection methods in the presence of a multipath fading channel with a single co-channel interferer as a function of various soft-bit data widths. It has been found that compressing the soft-bit data widths from the output of the equalizer to save memory can influence the likelihood decision of the coding scheme detect function and hence contribute to the overall performance loss of the system. Coding scheme detection errors can therefore force the channel decoder to either select the incorrect decoding scheme or have no clear decision which coding scheme to use resulting in the decoded radio block failing the block check sequence and contribute to the block error rate. For correct performance simulation, the performance of the full coding scheme detection must be taken into account.

Structures and negative thermal expansion properties of the one-dimensional cyanides, CuCN, AgCN and AuCN

The behaviour of the lattice parameters of HTCuCN (high-temperature form), AgCN and AuCN have been investigated as a function of temperature over the temperature range 90–490 K. All materials show one-dimensional negative thermal expansion (NTE) along the ––(M––CN)–– chain direction c (ac(HT-CuCN) ¼32.1 10–6 K1, ac(AgCN)¼23.910–6 K1 and ac(AuCN) ¼9.3106 K1 over the temperature range 90–490 K). The origin of this behaviour has been studied using RMC modelling of Bragg and total neutron diffraction data from AgCN and AuCN at 10 and 300 K. These analyses yield details of the local motions within the chains responsible for NTE. The low-temperature form of CuCN, LT-CuCN, has been studied using single-crystal X-ray diffraction. In this form of CuCN, wavelike distortions of the ––(Cu––CN)–– chains occur in the static structure, which are reminiscent of the motions seen in the RMC modelling of AgCN and AuCN, which are responsible for the NTE behaviour.

Technologies for biological containment of GM and Non-GM crops

International Perspective The development of GM technology continues to expand into increasing numbers of crops and conferred traits. Inevitably, the focus remains on the major field crops of soybean, maize, cotton, oilseed rape and potato with introduced genes conferring herbicide tolerance and/or pest resistance. Although there are comparatively few GM crops that have been commercialised to date, GM versions of 172 plant species have been grown in field trials in 31 countries. European Crops with Containment Issues Of the 20 main crops in the EU there are four for which GM varieties are commercially available (cotton, maize for animal feed and forage, and oilseed rape). Fourteen have GM varieties in field trials (bread wheat, barley, durum wheat, sunflower, oats, potatoes, sugar beet, grapes, alfalfa, olives, field peas, clover, apples, rice) and two have GM varieties still in development (rye, triticale). Many of these crops have hybridisation potential with wild and weedy relatives in the European flora (bread wheat, barley, oilseed rape, durum wheat, oats, sugar beet and grapes), with escapes (sunflower); and all have potential to cross-pollinate fields non-GM crops. Several fodder crops, forestry trees, grasses and ornamentals have varieties in field trials and these too may hybridise with wild relatives in the European flora (alfalfa, clover, lupin, silver birch, sweet chestnut, Norway spruce, Scots pine, poplar, elm, Agrostis canina, A. stolonifera, Festuca arundinacea, Lolium perenne, L. multiflorum, statice and rose). All these crops will require containment strategies to be in place if it is deemed necessary to prevent transgene movement to wild relatives and non-GM crops. Current Containment Strategies A wide variety of GM containment strategies are currently under development, with a particular focus on crops expressing pharmaceutical products. Physical containment in greenhouses and growth rooms is suitable for some crops (tomatoes, lettuce) and for research purposes. Aquatic bioreactors of some non-crop species (algae, moss, and duckweed) expressing pharmaceutical products have been adopted by some biotechnology companies. There are obvious limitations of the scale of physical containment strategies, addressed in part by the development of large underground facilities in the US and Canada. The additional resources required to grow plants underground incurs high costs that in the long term may negate any advantage of GM for commercial productioNatural genetic containment has been adopted by some companies through the selection of either non-food/feed crops (algae, moss, duckweed) as bio-pharming platforms or organisms with no wild relatives present in the local flora (safflower in the Americas). The expression of pharmaceutical products in leafy crops (tobacco, alfalfa, lettuce, spinach) enables growth and harvesting prior to and in the absence of flowering. Transgenically controlled containment strategies range in their approach and degree of development. Plastid transformation is relatively well developed but is not suited to all traits or crops and does not offer complete containment. Male sterility is well developed across a range of plants but has limitations in its application for fruit/seed bearing crops. It has been adopted in some commercial lines of oilseed rape despite not preventing escape via seed. Conditional lethality can be used to prevent flowering or seed development following the application of a chemical inducer, but requires 100% induction of the trait and sufficient application of the inducer to all plants. Equally, inducible expression of the GM trait requires equally stringent application conditions. Such a method will contain the trait but will allow the escape of a non-functioning transgene. Seed lethality (‘terminator’ technology) is the only strategy at present that prevents transgene movement via seed, but due to public opinion against the concept it has never been trialled in the field and is no longer under commercial development. Methods to control flowering and fruit development such as apomixis and cleistogamy will prevent crop-to-wild and wild-to-crop pollination, but in nature both of these strategies are complex and leaky. None of the genes controlling these traits have as yet been identified or characterised and therefore have not been transgenically introduced into crop species. Neither of these strategies will prevent transgene escape via seed and any feral apomicts that form are arguably more likely to become invasives. Transgene mitigation reduces the fitness of initial hybrids and so prevents stable introgression of transgenes into wild populations. However, it does not prevent initial formation of hybrids or spread to non-GM crops. Such strategies could be detrimental to wild populations and have not yet been demonstrated in the field. Similarly, auxotrophy prevents persistence of escapes and hybrids containing the transgene in an uncontrolled environment, but does not prevent transgene movement from the crop. Recoverable block of function, intein trans-splicing and transgene excision all use recombinases to modify the transgene in planta either to induce expression or to prevent it. All require optimal conditions and 100% accuracy to function and none have been tested under field conditions as yet. All will contain the GM trait but all will allow some non-native DNA to escape to wild populations or to non-GM crops. There are particular issues with GM trees and grasses as both are largely undomesticated, wind pollinated and perennial, thus providing many opportunities for hybridisation. Some species of both trees and grass are also capable of vegetative propagation without sexual reproduction. There are additional concerns regarding the weedy nature of many grass species and the long-term stability of GM traits across the life span of trees. Transgene stability and conferred sterility are difficult to trial in trees as most field trials are only conducted during the juvenile phase of tree growth. Bio-pharming of pharmaceutical and industrial compounds in plants Bio-pharming of pharmaceutical and industrial compounds in plants offers an attractive alternative to mammalian-based pharmaceutical and vaccine production. Several plantbased products are already on the market (Prodigene’s avidin, β-glucuronidase, trypsin generated in GM maize; Ventria’s lactoferrin generated in GM rice). Numerous products are in clinical trials (collagen, antibodies against tooth decay and non-Hodgkin’s lymphoma from tobacco; human gastric lipase, therapeutic enzymes, dietary supplements from maize; Hepatitis B and Norwalk virus vaccines from potato; rabies vaccines from spinach; dietary supplements from Arabidopsis). The initial production platforms for plant-based pharmaceuticals were selected from conventional crops, largely because an established knowledge base already existed. Tobacco and other leafy crops such as alfalfa, lettuce and spinach are widely used as leaves can be harvested and no flowering is required. Many of these crops can be grown in contained greenhouses. Potato is also widely used and can also be grown in contained conditions. The introduction of morphological markers may aid in the recognition and traceability of crops expressing pharmaceutical products. Plant cells or plant parts may be transformed and maintained in culture to produce recombinant products in a contained environment. Plant cells in suspension or in vitro, roots, root cells and guttation fluid from leaves may be engineered to secrete proteins that may be harvested in a continuous, non-destructive manner. Most strategies in this category remain developmental and have not been commercially adopted at present. Transient expression produces GM compounds from non-GM plants via the utilisation of bacterial or viral vectors. These vectors introduce the trait into specific tissues of whole plants or plant parts, but do not insert them into the heritable genome. There are some limitations of scale and the field release of such crops will require the regulation of the vector. However, several companies have several transiently expressed products in clinical and pre-clinical trials from crops raised in physical containment.

A mixture of trans-galactooligosaccharides reduces markers of metabolic syndrome and modulates the fecal microbiota and immune function of overweight adults

Metabolic syndrome is a set of disorders that increases the risk of developing cardiovascular disease. The gut microbiota is altered toward a less beneficial composition in overweight adults and this change can be accompanied by inflammation. Prebiotics such as galactooligosaccharides can positively modify the gut microbiota and immune system; some may also reduce blood lipids. We assessed the effect of a galactooligosaccharide mixture [Bi2 muno (B-GOS)] on markers of metabolic syndrome, gut microbiota, and immune function in 45 overweight adults with $3 risk factors associated with metabolic syndrome in a double-blind, randomized, placebo (maltodextrin)-controlled, crossover study (with a 4-wk wash-out period between interventions). Whole blood, saliva, feces, and anthropometric measurements were taken at the beginning, wk 6, and end of each 12-wk intervention period. Predominant groups of fecal bacteria were quantified and full blood count, markers of inflammation and lipid metabolism, insulin, and glucose were measured. B-GOS increased the number of fecal bifidobacteria at the expense of less desirable groups of bacteria. Increases in fecal secretory IgA and decreases in fecal calprotectin, plasma C-reactive protein, insulin, total cholesterol (TC), TG, and the TC:HDL cholesterol ratio were also observed. Administration of B-GOS to overweight adults resulted in positive effects on the composition of the gut microbiota, the immune response, and insulin, TC, and TG concentrations. B-GOSmay be a useful candidate for the enhancement of gastrointestinal health, immune function, and the reduction of metabolic syndrome risk factors in overweight adults.

Use of synthetic CaV2.2 Ca2+ channel peptides to study presynaptic function

Small, synthetic peptides based on specific regions of voltage-gated Ca2+ channels (VGCCs) have been widely used to study Ca2+ channel function and have been instrumental in confirming the contribution of specific amino acid sequences to interactions with putative binding partners. In particular, peptides based on the Ca2+ channel Alpha Interaction Domain (AID) on the intracellular region connecting domains I and II (the I-II loop) and the SYNaptic PRotein INTerction (synprint) site on the II-III loop have been widely used. Emerging evidence suggests that such peptides may themselves possess inherent functionality, a property that may be exploitable for future drug design. Here, we review our recent work using synthetic Ca2+ channel peptides based on sequences within the CaV2.2 amino terminal and I-II loop, originally identified as molecular determinates for G protein modulation, and their effects on VGCC function. These CaV2.2 peptides act as inhibitory modules to decrease Ca2+ influx with direct effects on VGCC gating, ultimately leading to a reduction of synaptic transmission. CaV2.2 peptides also attenuate G protein modulation of VGCCs. Amino acid substitutions generate CaV2.2 peptides with increased or decreased inhibitory effects suggesting that synthetic peptides can be used to further probe VGCC function and, potentially, form the basis for novel therapeutic development.

Genetic, structural, and molecular insights into the function of ras of complex proteins domains

Ras of complex proteins (ROC) domains were identified in 2003 as GTP binding modules in large multidomain proteins from Dictyostelium discoideum. Research into the function of these domains exploded with their identification in a number of proteins linked to human disease, including leucine-rich repeat kinase 2 (LRRK2) and death-associated protein kinase 1 (DAPK1) in Parkinson’s disease and cancer, respectively. This surge in research has resulted in a growing body of data revealing the role that ROC domains play in regulating protein function and signaling pathways. In this review, recent advances in the structural informa- tion available for proteins containing ROC domains, along with insights into enzymatic function and the integration of ROC domains as molecular switches in a cellular and organismal context, are explored.

Fruits, vegetables, 100% juice, and cognitive function

Although reviews of the association between polyphenol intake and cognition exist, research examining the cognitive effects of fruit, vegetable, and juice consumption across epidemiological and intervention studies has not been previously examined. Critical inclusion criteria were human participants, a measure of fruit, vegetable, or 100% juice consumption, an objective measure of cognitive function, and a clinical diagnosis of neuropsychological disease. Studies were excluded if consumption of fruit, vegetables, or juice was not assessed in isolation from other foods groups, or if there was no statistical control for education or IQ. Seventeen of 19 epidemiological studies and 3 of 6 intervention studies reported significant benefits of fruit, vegetable, or juice consumption for cognitive performance. The data suggest that chronic consumption of fruits, vegetables, and juices is beneficial for cognition in healthy older adults. The limited data from acute interventions indicate that consumption of fruit juices can have immediate benefits for memory function in adults with mild cognitive impairment; however, as of yet, acute benefits have not been observed in healthy adults. Conclusions regarding an optimum dietary intake for fruits, vegetables, and juices are difficult to quantify because of substantial heterogeneity in the categorization of consumption of these foods.