Support vector machine classification for large data sets via minimum enclosing ball clustering

Support vector machine (SVM) is a powerful technique for data classification. Despite of its good theoretic foundations and high classification accuracy, normal SVM is not suitable for classification of large data sets, because the training complexity of SVM is highly dependent on the size of data set. This paper presents a novel SVM classification approach for large data sets by using minimum enclosing ball clustering. After the training data are partitioned by the proposed clustering method, the centers of the clusters are used for the first time SVM classification. Then we use the clusters whose centers are support vectors or those clusters which have different classes to perform the second time SVM classification. In this stage most data are removed. Several experimental results show that the approach proposed in this paper has good classification accuracy compared with classic SVM while the training is significantly faster than several other SVM classifiers.

Feasibility of a Clinical Chemical Analysis Approach to Predict Misuse of Growth Promoting Hormones in Cattle

A study was performed to determine if targeted metabolic profiling of cattle sera could be used to establish a predictive tool for identifying hormone misuse in cattle. Metabolites were assayed in heifers (n ) 5) treated with nortestosterone decanoate (0.85 mg/kg body weight), untreated heifers (n ) 5), steers (n ) 5) treated with oestradiol benzoate (0.15 mg/kg body weight) and untreated steers (n ) 5). Treatments were administered on days 0, 14, and 28 throughout a 42 day study period. Two support vector machines (SVMs) were trained, respectively, from heifer and steer data to identify hormonetreated animals. Performance of both SVM classifiers were evaluated by sensitivity and specificity of treatment prediction. The SVM trained on steer data achieved 97.33% sensitivity and 93.85% specificity while the one on heifer data achieved 94.67% sensitivity and 87.69% specificity. Solutions of SVM classifiers were further exploited to determine those days when classification accuracy of the SVM was most reliable. For heifers and steers, days 17-35 were determined to be the most selective. In summary, bioinformatics applied to targeted metabolic profiles generated from standard clinical chemistry analyses, has yielded an accurate, inexpensive, high-throughput test for predicting steroid abuse in cattle.

Assisted Diagnosis of Cervical Intraepithelial Neoplasia (CIN)

This paper introduces an automated computer- assisted system for the diagnosis of cervical intraepithelial neoplasia (CIN) using ultra-large cervical histological digital slides. The system contains two parts: the segmentation of squamous epithelium and the diagnosis of CIN. For the segmentation, to reduce processing time, a multiresolution method is developed. The squamous epithelium layer is first segmented at a low (2X) resolution. The boundaries are further fine tuned at a higher (20X) resolution. The block-based segmentation method uses robust texture feature vectors in combination with support vector machines (SVMs) to perform classification. Medical rules are finally applied. In testing, segmentation using 31 digital slides achieves 94.25% accuracy. For the diagnosis of CIN, changes in nuclei structure and morphology along lines perpendicular to the main axis of the squamous epithelium are quantified and classified. Using multi-category SVM, perpendicular lines are classified into Normal, CIN I, CIN II, and CIN III. The robustness of the system in term of regional diagnosis is measured against pathologists' diagnoses and inter-observer variability between two pathologists is considered. Initial results suggest that the system has potential as a tool both to assist in pathologists' diagnoses, and in training.

A hierarchical multiclass support vector machine incorporated with holistic triple learning units

This paper proposes a new hierarchical learning structure, namely the holistic triple learning (HTL), for extending the binary support vector machine (SVM) to multi-classification problems. For an N-class problem, a HTL constructs a decision tree up to a depth of A leaf node of the decision tree is allowed to be placed with a holistic triple learning unit whose generalisation abilities are assessed and approved. Meanwhile, the remaining nodes in the decision tree each accommodate a standard binary SVM classifier. The holistic triple classifier is a regression model trained on three classes, whose training algorithm is originated from a recently proposed implementation technique, namely the least-squares support vector machine (LS-SVM). A major novelty with the holistic triple classifier is the reduced number of support vectors in the solution. For the resultant HTL-SVM, an upper bound of the generalisation error can be obtained. The time complexity of training the HTL-SVM is analysed, and is shown to be comparable to that of training the one-versus-one (1-vs.-1) SVM, particularly on small-scale datasets. Empirical studies show that the proposed HTL-SVM achieves competitive classification accuracy with a reduced number of support vectors compared to the popular 1-vs-1 alternative.

An efficient feature selection method for mobile devices with application to activity recognition

This paper presents a feature selection method for data classification, which combines a model-based variable selection technique and a fast two-stage subset selection algorithm. The relationship between a specified (and complete) set of candidate features and the class label is modelled using a non-linear full regression model which is linear-in-the-parameters. The performance of a sub-model measured by the sum of the squared-errors (SSE) is used to score the informativeness of the subset of features involved in the sub-model. The two-stage subset selection algorithm approaches a solution sub-model with the SSE being locally minimized. The features involved in the solution sub-model are selected as inputs to support vector machines (SVMs) for classification. The memory requirement of this algorithm is independent of the number of training patterns. This property makes this method suitable for applications executed in mobile devices where physical RAM memory is very limited. An application was developed for activity recognition, which implements the proposed feature selection algorithm and an SVM training procedure. Experiments are carried out with the application running on a PDA for human activity recognition using accelerometer data. A comparison with an information gain based feature selection method demonstrates the effectiveness and efficiency of the proposed algorithm.

The effects of temperature on the performance of electrochemical double layer capacitors

An electrochemical double layer capacitor test cell containing activated carbon xerogel electrodes and ionic liquid electrolyte was tested at 15, 25 and 40 OC to examine the effect of temperature on electrolyte resistance (RS) and equivalent series resistance (ESR) measured using impedance spectroscopy and capacitance using charge/discharge cycling. A commercial 10F capacitor was used as a comparison. Viscosity, ionic self-diffusion coefficients and differential scanning calorimetry measurements were used to provide an insight into the behaviour of the 1,2-dimethyl-3-propylimdazolium electrolyte. Both RS and ESR decreased with increasing temperature for both capacitors. Increasing the temperature also increased the capacitance for both the test cell and the commercial capacitor but proportionally more for the test cell. An increase in temperature decreased the ionic liquid electrolyte viscosity and increased the self diffusion coefficients of both the anion and the cation indicating an increase in dissociation and increase in ionic mobility.

Characterisation of protein stability in rod-insert vaginal rings

A major goal in vaccine development is elimination of the ‘cold chain’, the transport and storage system for maintenance and distribution of the vaccine product. This is particularly pertinent to liquid formulation of vaccines. We have previously described the rod-insert vaginal ring (RiR) device, comprising an elastomeric body into which are inserted lyophilised, rod-shaped, solid drug dosage forms, and having potential for sustained mucosal delivery of biomacromolecules, such as HIV envelope protein-based vaccine candidates. Given the solid, lyophilised nature of these insert dosage forms, we hypothesised that antigen stability may be significantly increased compared with more conventional solubilised vaginal gel format. In this study, we prepared and tested vaginal ring devices fitted with lyophilised rod inserts containing the model antigen bovine serum albumin (BSA). Both the RiRs and the gels that were freeze-dried to prepare the inserts were evaluated for BSA stability using PAGE, turbidimetry, microbial load, MALDI-TOF and qualitative precipitate solubility measurements. When stored at 4 oC, but not when stored at 40 oC / 75% RH, the RiR formulation offered protection against structural and conformational changes to BSA. The insert also retained matrix integrity and release characteristics. The results demonstrate that lypophilised gels can provide relative protection against degradation at lower temperatures compared to semi-solid gels. The major mechanism of degradation at 40 oC / 75% RH was shown to be protein aggregation. Finally, in a preliminary study, we found that addition of trehalose to the formulation significantly reduces the rate of BSA degradation as compared to the original formulation when stored at 40 oC /75% RH. Establishing the mechanism of degradation, and finding that degradation is decelerated in the presence of trehalose, will help inform further development of RiRs specifically and polymer based freeze-dried systems in general.

Improved conjugate gradient implementation for least squares support vector machines

As a promising method for pattern recognition and function estimation, least squares support vector machines (LS-SVM) express the training in terms of solving a linear system instead of a quadratic programming problem as for conventional support vector machines (SVM). In this paper, by using the information provided by the equality constraint, we transform the minimization problem with a single equality constraint in LS-SVM into an unconstrained minimization problem, then propose reduced formulations for LS-SVM. By introducing this transformation, the times of using conjugate gradient (CG) method, which is a greatly time-consuming step in obtaining the numerical solution, are reduced to one instead of two as proposed by Suykens et al. (1999). The comparison on computational speed of our method with the CG method proposed by Suykens et al. and the first order and second order SMO methods on several benchmark data sets shows a reduction of training time by up to 44%. (C) 2011 Elsevier B.V. All rights reserved.

Predicting organic acid concentration from UV/vis spectrometry measurements - A comparison of machine learning techniques

The concentration of organic acids in anaerobic digesters is one of the most critical parameters for monitoring and advanced control of anaerobic digestion processes. Thus, a reliable online-measurement system is absolutely necessary. A novel approach to obtaining these measurements indirectly and online using UV/vis spectroscopic probes, in conjunction with powerful pattern recognition methods, is presented in this paper. An UV/vis spectroscopic probe from S::CAN is used in combination with a custom-built dilution system to monitor the absorption of fully fermented sludge at a spectrum from 200 to 750 nm. Advanced pattern recognition methods are then used to map the non-linear relationship between measured absorption spectra to laboratory measurements of organic acid concentrations. Linear discriminant analysis, generalized discriminant analysis (GerDA), support vector machines (SVM), relevance vector machines, random forest and neural networks are investigated for this purpose and their performance compared. To validate the approach, online measurements have been taken at a full-scale 1.3-MW industrial biogas plant. Results show that whereas some of the methods considered do not yield satisfactory results, accurate prediction of organic acid concentration ranges can be obtained with both GerDA and SVM-based classifiers, with classification rates in excess of 87% achieved on test data.

Randomised controlled trial of ranitidine versus omeprazole in combination with antibiotics for eradication of Helicobacter pylori

This study compared high dose ranitidine versus low dose omeprazole with antibiotics for the eradication of H pylori. 80 patients (mean age 48 years, range 18-75) who had H pylori infection were randomised in an investigator-blind manner to either a two-week regime of omeprazole 20 mg daily, amoxycillin 500 mg tid and metronidazole 400 mg tid (OAM), or ranitidine 600 mg bd, amoxycillin 500 mg tid and metronidazole 400 mg tid (RAM), or omeprazole 20 mg daily and clarithromycin 500 mg tid (OC), or omeprazole 20 mg daily and placebo (OP). H pylori was eradicated in 6 of 19 patients in the OAM group (32%); 8 of 18 in the RAM group (44%), 4 of 15 in the OC group (27%); none of 18 in the OP group (0%). [<P0.005 for OAM, RAM, OC vs OP; P = N.S. between OAM, RAM, OC]. Overall metronidazole resistance was unexpectedly high at 58%. Eradication rates in metronidazole sensitive patients were 71% (5/7) and 100% (3/3) for OAM and RAM respectively. In conclusion, H pylori eradication rates using high dose ranitidine plus amoxycillin and metronidazole may be similar to that of low dose omeprazole in combination with the same antibiotics for omeprazole with clarithromycin. Overall eradication rates were low due to a high incidence of metronidazole resistance but were higher in metronidazole-sensitive patients. Even high dose ranitidine with two antibiotics achieves a relatively low eradication rate. These metronidazole-based regimens cannot be recommended in areas with a high incidence of metronidazole resistance.

A sequential algorithm for sparse support vector classifiers

Support vector machines (SVMs), though accurate, are not preferred in applications requiring high classification speed or when deployed in systems of limited computational resources, due to the large number of support vectors involved in the model. To overcome this problem we have devised a primal SVM method with the following properties: (1) it solves for the SVM representation without the need to invoke the representer theorem, (2) forward and backward selections are combined to approach the final globally optimal solution, and (3) a criterion is introduced for identification of support vectors leading to a much reduced support vector set. In addition to introducing this method the paper analyzes the complexity of the algorithm and presents test results on three public benchmark problems and a human activity recognition application. These applications demonstrate the effectiveness and efficiency of the proposed algorithm.


--------------------------------------------------------------------------------

Codeposition of organic carbon and arsenic in Bengal Delta aquifers

We present data showing that arsenic (As) was codeposited with organic carbon (OC) in Bengal Delta sediments as As and OC concentrations are highly (p <0.001) positively correlated in core profiles collected from widely dispersed geographical sites with different sedimentary depositional histories. Analysis of modern day depositional environments revealed that the As-OC correlations observed in cores are due to As retention and high OC inputs in vegetated zones of the deltaic environment. We hypothesize that elevated concentrations of As occur in vegetated wetland sediments due to concentration and retention of arsenate in aerated root zones and animal burrows where copious iron(III) oxides are deposited. On burial of the sediment, degradation of organic carbon from plant and animal biomass detritus provides the reducing conditions to dissolve iron(III) oxides and release arsenite into the porewater. As tubewell abstracted aquifer water is an invaluable resource on which much of Southeast Asia is now dependent, this increased understanding of the processes responsible for As buildup and release will identify, through knowledge of the palaeosedimentary environment, which sediments are at most risk of having high arsenic concentrations in porewater. Our data allow the development of a new unifying hypothesis of how As is mobilized into groundwaters in river flood plains and deltas of Southeast Asia, namely that in these highly biologically productive environments, As and OC are codeposited, and the codeposited OC drives As release from the sediments.

Premorbid IQ in patients with functional psychosis and their first-degree relatives

Numerous studies have found deficits in premorbid IQ in schizophrenic patients, but it is not clear whether this deficit is shared by (a) patients with other functional psychoses, and (b) relatives of these patients. Ninety-one schizophrenic patients, 66 affective psychotic patients (29 schizoaffective and 37 manic or depressed), and 50 normal control subjects were administered the National Adult Reading Test (NART) which provides an estimate of premorbid IQ. The NART was also completed by 85 first-degree relatives of schizophrenic patients and by 65 first-degree relatives of affective psychotic patients. After adjustments were made for sex, social class, ethnicity and years of education, schizophrenic patients had significantly lower premorbid IQ than their relatives, the affective psychotic patients and controls. Manic and depressed patients had significantly lower NART scores than their first-degree relatives, but schizoaffective patients did not, and neither group differed significantly from controls. There was no significant difference in premorbid IQ between patients who had experienced obstetric complications (OC +) and those who had not (OC -). Both OC + and OC - schizophrenic patients differed significantly from their relatives, but the disparity was greatest between OC + patients and their relatives. Relatives of OC + schizophrenic patients had significantly higher IQ than relatives of OC - schizophrenic patients. (C) 2000 Elsevier Science B.V. All rights reserved.

Characterization of the six glycosyltransferases involved in the biosynthesis of Yersinia enterocolitica serotype O:3 lipopolysaccharide outer core

Yersinia enterocolitica (Ye) is a gram-negative bacterium; Ye serotype O:3 expresses lipopolysaccharide (LPS) with a hexasaccharide branch known as the outer core (OC). The OC is important for the resistance of the bacterium to cationic antimicrobial peptides and also functions as a receptor for bacteriophage phiR1-37 and enterocoliticin. The biosynthesis of the OC hexasaccharide is directed by the OC gene cluster that contains nine genes (wzx, wbcKLMNOPQ, and gne). In this study, we inactivated the six OC genes predicted to encode glycosyltransferases (GTase) one by one by nonpolar mutations to assign functions to their gene products. The mutants expressed no OC or truncated OC oligosaccharides of different lengths. The truncated OC oligosaccharides revealed that the minimum structural requirements for the interactions of OC with bacteriophage phiR1-37, enterocoliticin, and OC-specific monoclonal antibody 2B5 were different. Furthermore, using chemical and structural analyses of the mutant LPSs, we could assign specific functions to all six GTases and also revealed the exact order in which the transferases build the hexasaccharide. Comparative modeling of the catalytic sites of glucosyltransferases WbcK and WbcL followed by site-directed mutagenesis allowed us to identify Asp-182 and Glu-181, respectively, as catalytic base residues of these two GTases. In general, conclusive evidence for specific GTase functions have been rare due to difficulties in accessibility of the appropriate donors and acceptors; however, in this work we were able to utilize the structural analysis of LPS to get direct experimental evidence for five different GTase specificities.

Characterization and biological role of the O-polysaccharide gene cluster of Yersinia enterocolitica serotype O:9

Yersinia enterocolitica serotype O:9 is a gram-negative enteropathogen that infects animals and humans. The role of lipopolysaccharide (LPS) in Y. enterocolitica O:9 pathogenesis, however, remains unclear. The O:9 LPS consists of lipid A to which is linked the inner core oligosaccharide, serving as an attachment site for both the outer core (OC) hexasaccharide and the O-polysaccharide (OPS; a homopolymer of N-formylperosamine). In this work, we cloned the OPS gene cluster of O:9 and identified 12 genes organized into four operons upstream of the gnd gene. Ten genes were predicted to encode glycosyltransferases, the ATP-binding cassette polysaccharide translocators, or enzymes required for the biosynthesis of GDP-N-formylperosamine. The two remaining genes within the OPS gene cluster, galF and galU, were not ascribed a clear function in OPS biosynthesis; however, the latter gene appeared to be essential for O:9. The biological functions of O:9 OPS and OC were studied using isogenic mutants lacking one or both of these LPS parts. We showed that OPS and OC confer resistance to human complement and polymyxin B; the OPS effect on polymyxin B resistance could be observed only in the absence of OC.