Mixed-Precision Kernel Recursive Least Squares.

Kernel recursive least squares (KRLS) is a widely used online machine learning algorithm for time series predictions. In this article, we present the mixed-precision KRLS, producing equivalent prediction accuracy to double-precision KRLS with a higher training throughput and a lower memory footprint. The mixed-precision KRLS applies single-precision arithmetic to the computation components being not only numerically resilient but also computationally intensive. Our mixed-precision KRLS demonstrates the 1.32, 1.15, 1.29, 1.09, and 1.08× training throughput improvements using 24.95%, 24.74%, 24.89%, 24.48%, and 24.20% less memory footprint without losing any prediction accuracy compared to double-precision KRLS for a 3-D nonlinear regression, a Lorenz chaotic time series, a Mackey-Glass chaotic time series, a sunspot number time series, and a sea surface temperature time series, respectively.

Expediting assessments of database performance for streams of respiratory parameters.

A new methodology is proposed to compare database performance for streams of patient respiratory data from patients in an intensive care unit. New metrics are proposed through which databases may be compared both for this and similar streaming applications in the domain of the Internet of Things. Studies are reported using simulated patient data for four freely available databases. The statistical technique of non-parametric bootstrapping is used to minimise the total running time of the tests. We report mean values and bias corrected and accelerated confidence intervals for each metric and use these to compare the databases. We find that, among the four databases tested, ScaleDB is an optimum database technology when handling between 200 and 800 patients in this application, while PostgreSQL performs best outside of this range. Comparing the non-parametric bootstrapping method to a complete set of tests shows that the two approaches give results differing by a few percent.

A real time metabolomic profiling approach to detecting fish fraud using rapid evaporative ionisation mass spectrometry.

INTRODUCTION: Fish fraud detection is mainly carried out using a genomic profiling approach requiring long and complex sample preparations and assay running times. Rapid evaporative ionisation mass spectrometry (REIMS) can circumvent these issues without sacrificing a loss in the quality of results. OBJECTIVES: To demonstrate that REIMS can be used as a fast profiling technique capable of achieving accurate species identification without the need for any sample preparation. Additionally, we wanted to demonstrate that other aspects of fish fraud other than speciation are detectable using REIMS. METHODS: 478 samples of five different white fish species were subjected to REIMS analysis using an electrosurgical knife. Each sample was cut 8-12 times with each one lasting 3-5 s and chemometric models were generated based on the mass range m/z 600-950 of each sample. RESULTS: The identification of 99 validation samples provided a 98.99% correct classification in which species identification was obtained near-instantaneously (≈ 2 s) unlike any other form of food fraud analysis. Significant time comparisons between REIMS and polymerase chain reaction (PCR) were observed when analysing 6 mislabelled samples demonstrating how REIMS can be used as a complimentary technique to detect fish fraud. Additionally, we have demonstrated that the catch method of fish products is capable of detection using REIMS, a concept never previously reported. CONCLUSIONS: REIMS has been proven to be an innovative technique to help aid the detection of fish fraud and has the potential to be utilised by fisheries to conduct their own quality control (QC) checks for fast accurate results.