Kernel Processing Research Articles

Nondeterministic Kriging for Engineering Design Exploration

In this paper, the nondeterministic kriging (NDK) method is proposed, aiming for the applications of engineering design exploration, especially when only a limited number of random samples is available from either nondeterministic simulations or physical experiments under uncertainty. To handle nondeterministic data, the proposed NDK method uses separate aleatory and epistemic uncertainty processes. In a general situation in which resources are limited in generating random samples, an aleatory variance is assessed via a local regression kernel process. It is often found that a prediction model built with a conventional kriging suffers from the overfitting issue, which becomes worse with noisy and random data. The proposed NDK method can provide physically meaningful insights into both the main trend and the prediction uncertainty of system behaviors by capturing uncertainty in the sample data and suppressing the numerical instability. The predicted uncertainty from the proposed approach can be represented in terms of distinguishable aleatory and epistemic uncertainties, which will be useful in a decision-making process for an adaptive model building and design exploration. The potential benefits of using the proposed NDK method are demonstrated with multiple numerical examples, including mathematical and aircraft concept design problems.

Handwriting recognition system based on FPGA

An FPGA-based handwriting recognition system is proposed in this paper, which uses convolutional neural network algorithm to realize the recognition of MNIST digital sets. Hardware design techniques including image cropping, convolution, activation functions, pooling, pipeline processing, and parallel processing of multiple convolution kernels are described. Using the characteristics of parallel computing of hardware circuits, the processing speed of the MNIST digital set detection system is accelerated. The proposed MNIST detection architecture is designed using the Verilog HDL and implemented in the Altera DE2 FPGA development board.

Design and Development of Roasted Breadfruit Kernel Processing Machine

An integrated machine has been designed and developed to aid mass production of unparboiled roasted breadfruit kernel in order to avert persistent scarcity of this unique nutritious snack. The machine was fabricated using local sourced standard materials with electric motor, roaster, dehuller, blower, kernel and shell discharging chutes as its major components. Performance analysis revealed that this system eliminated human contact with the breadfruit nuts during processing while 0.74kg/h, 93.63% and 167.93kJ/kg constitute its throughput, efficiency and specific energy consumption respectively. Thus, adoption of this innovation is recommended as a veritable option for quality roasted breadfruit kernel production since it reduced drudgery and improved hygiene.

Compressive Resistance of Groundnut Kernels as Influenced by Kernel Size

Mechanical properties of agricultural materials are essential for the proper design and fabrication of harvesting, handling, and processing equipment/machineries. In this research, some compressive resistance (force, energy and deformation at rupture point) of two groundnut (SAMNUT 10 and SAMNUT 11) kernels were investigated in terms of kernel sizes. During the test, the groundnut kernels were loaded quasi-statically in the axial orientation at a compressive loading rate of 20 mm/min, using the Universal Testing Machine. Results obtained from the test showed that kernel size and groundnut variety had significant (P ≤0.05) effect on all the mechanical parameters studied. The force required for initiating the kernel rupture increased from 37.21 to 76.10 N for SAMNUT 10; and 30.10 to 64.19 N for SAMNUT 11, as the kernels size increased from small to large size. In addition, the energy absorbed by the kernel at rupture point increased from 0.021 to 0.054 Nm for SAMNUT 10; and 0.016 to 0.044 Nm for SAMNUT 11, for the small and large kernel sizes respectively. Furthermore, the results showed that SAMNUT 10 kernels had slightly higher compressive resistance values than the SAMNUT 11 kernels. Data obtained from this research will help to design and fabricate equipment used in handling and processing of groundnut kernels.

Aqueous two-phase system for the extraction of amygdalin from the debitterized water of apricot kernels

ABSTRACTAqueous two-phase system (ATPS) is regarded as an effective and eco-friendly technique to extract some bioactive compounds. In order to promote the added value and reduce the discharge in the industrial processing of apricot kernels, the aqueous two-phase extraction conditions were optimized with the evaluation of the distribution coefficient (K), ratio of volume (R) and extraction yield (Y) to obtain a higher extraction of amygdalin from the debitterizing water concentrate (DWC) in this paper. The results indicate that the optimal conditions were as follows: 30% (w/w) ethanol and 20% (w/w) ammonium sulfate of the ATPS, at 35°C and pH of 7. Under these conditions, the highest extraction yield was 90.37% and the purity could be increased by more than two times. In a word, the ATPS can be considered as an effective method to recycle the amygdalin from the DWC in the apricot kernels processing.

Gaussian Process Based Model-free Control with Q-Learning

Abstract The aim of this paper is to demonstrate a new algorithm for Machine Learning (ML) based on Gaussian Process Regression (GPR) and how it can be used as a practical control design technique. An optimized control law for a nonlinear process is found directly by training the algorithm on noisy data collected from the process when controlled by a sub-optimal controller. A simplified nonlinear Fan Coil Unit (FCU) model is used as an example for which the fan speed control is designed using the off-policy Q-learning algorithm. Additionally, the algorithm properties are discussed, i.e. learning process robustness, Gaussian Process (GP) kernel functions choice. The simulation results are compared to a simple PI design based on a linearized model.

Refinement-Based Specification and Security Analysis of Separation Kernels

Assurance of information-flow security by formal methods is mandated in security certification of separation kernels. As an industrial standard for improving safety, ARINC 653 has been complied with by mainstream separation kernels. Due to the new trend of integrating safe and secure functionalities into one separation kernel, security analysis of ARINC 653 as well as a formal specification with security proofs are thus significant for the development and certification of ARINC 653 compliant S eparation K ernels (ARINC SKs). This paper presents a specification development and security analysis method for ARINC SKs based on refinement. We propose a generic security model and a stepwise refinement framework. Two levels of functional specification are developed by the refinement. A major part of separation kernel requirements in ARINC 653 are modeled, such as kernel initialization, two-level scheduling, partition and process management, and inter-partition communication. The formal specification and its security proofs are carried out in the Isabelle/HOL theorem prover. We have reviewed the source code of one industrial and two open-source ARINC SK implementations, i.e., VxWorks 653, XtratuM, and POK, in accordance with the formal specification. During the verification and code review, six security flaws, which can cause information leakage, are found in the ARINC 653 standard and the implementations.

Efficient Successive-Cancellation Polar Decoder Based on Redundant LLR Representation

This brief presents an efficient architecture of the polar decoder that employs the successive-cancellation (SC) decoding algorithm. In the SC decoding algorithm, each bit is decoded successively by recursively calculating the log likelihood ratio (LLR) based on two kernels. This brief proposes a novel LLR representation scheme so that the kernel processing can be realized in low-complexity and high-speed circuitry. A 1024-bit polar decoder was designed and implemented based on the proposed scheme using a ${0.18~\mu }\text{m}$ CMOS process. Its throughput is ${252R}$ Mb/s for the rate- ${R}$ code, and the gate count is 256K. By the proposed LLR representation scheme, the decoding speed is increased by 18% while the gate count is not increased when compared to the same decoder designed with the signed-magnitude scheme. In terms of the throughput efficiency, the proposed decoder is 1.34 times superior to the previous state-of-the-art decoder.

Efficient global optimization of constrained mixed variable problems

Due to the increasing demand for high performance and cost reduction within the framework of complex system design, numerical optimization of computationally costly problems is an increasingly popular topic in most engineering fields. In this paper, several variants of the Efficient Global Optimization algorithm for costly constrained problems depending simultaneously on continuous decision variables as well as on quantitative and/or qualitative discrete design parameters are proposed. The adaptation that is considered is based on a redefinition of the Gaussian Process kernel as a product between the standard continuous kernel and a second kernel representing the covariance between the discrete variable values. Several parameterizations of this discrete kernel, with their respective strengths and weaknesses, are discussed in this paper. The novel algorithms are tested on a number of analytical test-cases and an aerospace related design problem, and it is shown that they require fewer function evaluations in order to converge towards the neighborhoods of the problem optima when compared to more commonly used optimization algorithms.

Convolution neural network-based time-domain equalizer for DFT-Spread OFDM VLC system

Abstract This paper presents a novel time-domain equalizer for visible light communication (VLC) system using machine learning (ML) method. In this work, we employ discrete Fourier transform spread orthogonal frequency division multiplexing (DFT-S-OFDM) as modem scheme and convolution neural network (CNN) as kernel processing unit of equalizer. After estimating channel state information (CSI) from training sequence, the proposed equalizer recovers transmitted symbols according to the estimated CSI. Numerical simulations indicate that the equalizer can significantly enhance bit error rate (BER) performance. For example, when signal-to-noise ratio (SNR) is 20 dB and 16/32/64-quadrature amplitude modulation (QAM) is exploited, original BER is about 0.5 while the BER after recovery achieves 1 0 − 5 , which is much lower than forward error correction (FEC) limit 3 . 8 × 1 0 − 3 . This work promotes the application of ML in VLC domain. To the best of our knowledge, this is the first time a CNN-based equalizer has been explored.

Advanced Classification of Coffee Beans with Fatty Acids Profiling to Block Information Loss

Classification is a kernel process in the standardization, grading, and sensory aspects of coffee industries. The chemometric data of fatty acids and crude fat are used to characterize the varieties of coffee. Two category classifiers were used to distinguish the species and roasting degree of coffee beans. However, the fatty acid profiling with normalized data gave a bad discriminant result in the classification study with mixed dimensions in species and roasted degree. The result of the predictive model is in conflict with the context of human cognition, since roasted coffee beans are easily visually distinguished from green coffee beans. By exploring the effects of error analysis and information processing technologies, the lost information was identified as a bias–variance tradeoff derived from the percentile normalization. The roasting degree as extensive information was attenuated by the percentile normalization, but the cultivars as intensive information were enhanced. An informational spiking technique is proposed to patch the dataset and block the information loss. The identified blocking of informational loss could be available for multidimensional classification systems based on the chemometric data.

Development and verification of resonance elastic scattering kernel processing module in nuclear data processing code NECP-Atlas

Some efforts have been made to exactly consider the effect of neutron up-scattering caused by thermal motion of target nuclei and resonance elastic scattering on the multi-group cross sections and scattering matrices. Firstly, the resonance elastic scattering kernel (RESK) formulations for anisotropic scattering up to any Legendre order has been adopted to represent the exact Doppler broadened energy transfer kernels. A semi-analytical integration method is applied to perform the RESK calculations. Combining with the RESK calculations, a linearization algorithm is proposed to generate the RESK interpolation tables. The RESK data can be interpolated precisely based on the interpolation tables to reduce the calculation burden. Secondly, a neutron slowing-down equation solver is developed based on the RESK instead of the conventional asymptotic scattering kernel, where the effect of neutron up-scattering on the neutron energy spectrum can be exactly considered. The accuracy of the multi-group cross sections and scattering matrices are significantly improved when the exact energy spectrum is applied for the group collapsing calculations. All the methods mentioned above have been implemented into a newly developed nuclear data processing code called NECP-Atlas. Numerical results show that the proposed methods are capable of producing accurate multi-group cross sections for downstream calculations; the fuel Doppler coefficients and eigenvalues of the real problems will be improved if the up-scattering effect is incorporated into the multi-group cross sections and matrices.

Hyperspectral Recovery from RGB Images using Gaussian Processes

We propose to recover spectral details from RGB images of known spectral quantization by modeling natural spectra under Gaussian Processes and combining them with the RGB images. Our technique exploits Process Kernels to model the relative smoothness of reflectance spectra, and encourages non-negativity in the resulting signals for better estimation of the reflectance values. The Gaussian Processes are inferred in sets using clusters of spatio-spectrally correlated hyperspectral training patches. Each set is transformed to match the spectral quantization of the test RGB image. We extract overlapping patches from the RGB image and match them to the hyperspectral training patches by spectrally transforming the latter. The RGB patches are encoded over the transformed Gaussian Processes related to those hyperspectral patches and the resulting image is constructed by combining the codes with the original processes. Our approach infers the desired Gaussian Processes under a fully Bayesian model inspired by Beta-Bernoulli Process, for which we also present the inference procedure. A thorough evaluation using three hyperspectral datasets demonstrates the effective extraction of spectral details from RGB images by the proposed technique.

Collocating Local Volatility: A Competitive Alternative to Stochastic Local Volatility Models

We discuss a competitive alternative to stochastic local volatility models, namely the Collocating Volatility (CV) model, introduced in Grzelak (2016). The CV model consists of two elements, a 'kernel process' that can be efficiently evaluated and a local volatility function. The latter, based on stochastic collocation – e.g. Babuska et al. (2007), Witteveen et al. (2012) – connects the kernel process to the market and allows the CV model to be perfectly calibrated to European-type options. In this article we consider three different kernel process choices: the Ornstein-Uhlenbeck (OU) and Cox-Ingersoll-Ross (CIR) processes and the Heston model. The kernel process controls the forward smile and allows for an accurate and efficient calibration to exotic options, while the perfect calibration to liquid market quotes is preserved. We confirm this by numerical experiments, in which we calibrate the OU-CV, CIR-CV and Heston-CV models to FX barrier options.

KAPABILITAS PROSES DENGAN ESTIMASI FUNGSI DENSITAS KERNEL PADA PRODUKSI DENIM DI PT APAC INTI CORPORA

The quality of production becomes one of the basic factors of consumer decisions in choosing a product. Quality control is needed to control the production process. Control chart is a tool used in performing statistical quality control. One of the alternatives used when the data obtained is not known distribution is analyzed by nonparametric approach based on estimation of kernel density function. The most important thing in estimating kernel density function is optimal bandwidth selection (h) which minimizes Cross Validation (CV) value. Some of the kernel functions used in this research are Rectangular, Epanechnikov, Triangular, Biweight, and Gaussian. If the process control chart is statistically controlled, a process capability analysis can be calculated using the process conformity index to determine the nature of the process capability. In this research, the kernel control chart and process conformity index were used to analyze the slope shift of Akira-F style fabric and Corvus-SI style on the production of denim fabric at PT Apac Inti Corpora. The results of the analysis show that the production process for Akira-F style is statistically controlled, but Ypk > Yp is 0.889823 > 0,508059 indicating that the process is still not in accordance with the specified limits set by the company, while for Corvus- SI is statistically controlled and Ypk < Yp is 0.637742 < 0.638776 which indicates that the process is in accordance with the specification limits specified by the company. Keywords: kernel density function estimation, Cross Validation, kernel control chart, denim fabric, process capability

In situ rumen degradation of kernels from short-season corn silage hybrids as affected by processing.

The objective of this in situ study was to evaluate the rumen degradability of kernels from short-season corn hybrids grown for silage in Western Canada (Lacombe, AB) and determine whether decreasing kernel particle size would enhance ruminal degradability in a similar manner for all hybrids. The study was a completely randomized design with 3 beef cows (replicates) and a 6 (hybrid) × 3 (particle size) factorial arrangement of treatments. Kernels were processed to generate three different particle sizes: large (2.3 mm), medium (1.4 mm), and small (0.7 mm). Processed samples were incubated in the rumen for 0, 3, 6, 12, 24, and 48 h using the in situ method and degradation kinetics of DM and starch were determined. Effective rumen degradability (ED) was estimated using a passage rate of 0.04 (ED4), 0.06 (ED6), and 0.08/h (ED8). Hybrids exhibited a range in whole plant DM content (23.7 to 25.0%), starch content (15.9 to 28.1% DM), kernel hardness (21.9 to 34.4 s/20 g) and density (3.57 to 4.18 g/mL), and prolamin content (8.24 to 11.34 g/100 g starch). Differences in digestion kinetics among hybrids were generally more pronounced for starch than DM. The hybrids differed in starch degradability (P < 0.05), with earlier maturing hybrids having lower A fraction, lower k d, and lower ED, with hybrid effects on ED being accentuated with faster passage rate. Kernel DM content (r = -0.85, -0.87), hardness (r = -0.89, -0.86), and density (r = -0.84, -0.85) were negatively correlated with ED4 and ED8 of starch, respectively, due mainly to decreased k d of fraction B. Reducing the particle size of kernels increased ED of starch due to increased A fraction and k d of the B fraction. A tendency (P = 0.09) for hybrid × processing effects for ED6 and ED8 indicated that processing had greater effects on increasing ED of starch for earlier maturing hybrids. We conclude that short-season hybrids that mature early may have lower ED of DM and starch and would benefit from prolonged ensilage time. Kernel processing during silage making is recommended for short-season corn hybrids as a means of enhancing rumen availability of starch.

MGPfusion: predicting protein stability changes with Gaussian process kernel learning and data fusion.

MotivationProteins are commonly used by biochemical industry for numerous processes. Refining these proteins’ properties via mutations causes stability effects as well. Accurate computational method to predict how mutations affect protein stability is necessary to facilitate efficient protein design. However, accuracy of predictive models is ultimately constrained by the limited availability of experimental data.ResultsWe have developed mGPfusion, a novel Gaussian process (GP) method for predicting protein’s stability changes upon single and multiple mutations. This method complements the limited experimental data with large amounts of molecular simulation data. We introduce a Bayesian data fusion model that re-calibrates the experimental and in silico data sources and then learns a predictive GP model from the combined data. Our protein-specific model requires experimental data only regarding the protein of interest and performs well even with few experimental measurements. The mGPfusion models proteins by contact maps and infers the stability effects caused by mutations with a mixture of graph kernels. Our results show that mGPfusion outperforms state-of-the-art methods in predicting protein stability on a dataset of 15 different proteins and that incorporating molecular simulation data improves the model learning and prediction accuracy.Availability and implementationSoftware implementation and datasets are available at github.com/emmijokinen/mgpfusion.Supplementary information Supplementary data are available at Bioinformatics online.

Efficient nonparametricn-body force fields from machine learning

We provide a definition and explicit expressions for $n$-body Gaussian Process (GP) kernels which can learn any interatomic interaction occurring in a physical system, up to $n$-body contributions, for any value of $n$. The series is complete, as it can be shown that the "universal approximator" squared exponential kernel can be written as a sum of $n$-body kernels. These recipes enable the choice of optimally efficient force models for each target system, as confirmed by extensive testing on various materials. We furthermore describe how the $n$-body kernels can be "mapped" on equivalent representations that provide database-size-independent predictions and are thus crucially more efficient. We explicitly carry out this mapping procedure for the first non-trivial (3-body) kernel of the series, and show that this reproduces the GP-predicted forces with $\text{meV/} \AA$ accuracy while being orders of magnitude faster. These results open the way to using novel force models (here named "M-FFs") that are computationally as fast as their corresponding standard parametrised $n$-body force fields, while retaining the nonparametric character, the ease of training and validation, and the accuracy of the best recently proposed machine learning potentials.

Logistic model-based genetic analysis for kernel filling in a maize RIL population

Kernel filling is an important factor that directly affects kernel yield in maize. Based on a Logistic model, the process of kernel filling in maize can be effectively fitted, and the characteristic parameters with biological significance can be estimated. To clarify the genetic mechanism of characteristic parameters of kernel filling in maize, a recombinant inbred line (RIL) population including 208 lines derived from the maize inbred lines DH1M and T877 were evaluated in Nantong in 2015 and in Yangzhou in 2016, respectively. The kernel dry weights of recombinant inbred lines were measured 10, 15, 20, 25, 30, 35, 40, 43, 46, 49, 52, 55, 58 and 61 days after pollination (DAP). A total of 12 characteristic parameters related to kernel filling were estimated in different environments using the Logistic model. These parameters showed abundant phenotypic variation across two environments in the recombinant inbred line population. Some more ideal genotypes were selected through clustering based on BLUP values of characteristic parameters. Genetic analysis indicated that the 12 characteristic parameters conformed to the “major gene plus polygenes” model. The results of two environments were reproduced well. Most of the characteristic parameters related to kernel filling were controlled by two major genes, and a few characteristic parameters were controlled by three or four major genes. In addition, the genetic models of some characteristic parameters differed in the two environments due to interactions between the genes and environments. This study not only laid a foundation for further clarifying the genetic mechanism of maize kernel filling and mapping the related genes but also suggests a new paradigm for dynamic developing traits.

424 Effects of Kernel Processing of Corn Silage with and without the Brown Midrib Trait on Nutrient Digestion By Finishing Steers.

424 Effects of Kernel Processing of Corn Silage with and without the Brown Midrib Trait on Nutrient Digestion By Finishing Steers.