SVM Compound Kernel Functions for Vehicle Target Classification

The focus of this paper is to explore the use of kernel combinations of the support vector machines (SVMs) for vehicle classification. Being the primary component of the SVM, the kernel functions are responsible for the pattern analysis of the vehicle dataset and to bridge its linear and non-linear features. However, the choice of the type of kernel functions has characteristics and limitations that are highly dependent on the parameters. Thus, in order to overcome these limitations, a method of compounding kernel function for vehicle classification is hereby introduced and discussed. The vehicle classification accuracy of the compound kernel function presented is then compared to the accuracies of the conventional classifications obtained from the four commonly used individual kernel functions (linear, quadratic, cubic, and Gaussian functions). This study provides the following contributions: (1) The classification method is able to determine the rank in terms of accuracies of the four individual kernel functions; (2) The method is able to combine the top three individual kernel functions; and (3) The best combination of the compound kernel functions can be determined.

Simple sequence repeat-based identification of Canadian malting barley varieties

Perry, D. J., Fernando, U. and Lee, S-J. 2014. Simple sequence repeat-based identification of Canadian malting barley varieties. Can. J. Plant Sci. 94: 485–496. Practical and reliable means to identify barley varieties are required to provide assurances in segregated grain handling and for quality control in the malting and brewing industry. A set of 10 simple sequence repeat (SSR) markers was selected to differentiate among malting barley varieties grown in Canada. Modification of some PCR primers permitted assembly into two five-marker multiplexes that may be examined simultaneously using an electrophoresis-based DNA analyzer. These markers were surveyed in multiple individual kernels of each of 48 barley varieties grown in Canada, including 31 malting varieties and 17 popular feed varieties. Variation within varieties was common and three general categories of intra-variety polymorphism were recognized: (1) primary biotypes, which were characterized by a fairly even distribution of two alleles at one or more marker loci and complete mixture of allele combinations among the polymorphic loci; (2) uncommon, distinctly different variants; and (3) putative recent SSR mutations. Differentiation among varieties was complete with the exception of one pair of related six-row feed varieties (AC Rosser and AC Ranger) that was indistinguishable and one group of three very closely related two-row malting varieties (CDC Kendall, CDC PolarStar and Norman) that, on an individual-kernel basis, were only partially distinguishable using these markers. Simple, rapid individual-kernel DNA preparation methods were also developed for use in conjunction with the multiplexed markers to provide a convenient, effective and relatively inexpensive tool that may be used for barley variety identification, purity analysis or quantification of variety mixtures.

Ensemble Pruning for Data Dependant Learners

Ensemble pruning searches for a selective subset of members that performs as well as, or better than ensemble of all members. However, in the accuracy / diversity pruning framework, generalization ability of target ensemble is not considered, and moreover, there is not clear relationship between them. In this paper, we proof that ensemble formed by members of better generalization ability is also of better generalization ability. We adopt learning with both labeled and unlabeled data to improve generalization ability of member learners. A data dependant kernel determined by a set of unlabeled points is plugged in individual kernel learners to improve generalization ability, and ensemble pruning is launched as much previous work. The proposed method is suitable for both single-instance and multi-instance learning framework. Experimental results on 10 UCI data sets for single-instance learning and 4 data sets for multi-instance learning show that subensemble formed by the proposed method is effective.

Interlaboratory Study of DNA Extraction from Multiple Ground Samples, Multiplex Real-Time PCR, and Multiplex Qualitative PCR for Individual Kernel Detection System of Genetically Modified Maize

In many countries, the labeling of grains, feed, and foodstuff is mandatory if the genetically modified (GM) organism content exceeds a certain level of approved GM varieties. We previously developed an individual kernel detection system consisting of grinding individual kernels, DNA extraction from the individually ground kernels, GM detection using multiplex real-time PCR, and GM event detection using multiplex qualitative PCR to analyze the precise commingling level and varieties of GM maize in real sample grains. We performed the interlaboratory study of the DNA extraction with multiple ground samples, multiplex real-time PCR detection, and multiplex qualitative PCR detection to evaluate its applicability, practicality, and ruggedness for the individual kernel detection system of GM maize. DNA extraction with multiple ground samples, multiplex real-time PCR, and multiplex qualitative PCR were evaluated by five laboratories in Japan, and all results from these laboratories were consistent with the expected results in terms of the commingling level and event analysis. Thus, the DNA extraction with multiple ground samples, multiplex real-time PCR, and multiplex qualitative PCR for the individual kernel detection system is applicable and practicable in a laboratory to regulate the commingling level of GM maize grain for GM samples, including stacked GM maize.