HYBRID GREY RELATIONAL ARTIFICIAL NEURAL NETWORK AND AUTO REGRESSIVE INTEGRATED MOVING AVERAGE MODEL FOR FORECASTING TIME-SERIES DATA

The aim of this study is to develop a new hybrid model by combining a linear and nonlinear model for forecasting time-series data. The proposed model (GRANN_ARIMA) integrates nonlinear grey relational artificial neural network (GRANN) and a linear autoregressive integrated moving average (ARIMA) model by combining new features and grey relational analysis to select the appropriate inputs and hybridization succession. To validate the performance of the proposed model, small and large scale data sets are used. The forecasting performance is compared with several models, and these include: individual models (ARIMA, multiple regression, GRANN), several hybrid models (MARMA, MR_ANN, ARIMA_ANN), and an artificial neural network (ANN) trained using a Levenberg Marquardt algorithm. The experiments have shown that the proposed model has outperformed other models with 99.5% forecasting accuracy for small-scale data and 99.84% for large-scale data. The obtained empirical results have proven that the GRANN_ARIMA model can provide a better alternative for time-series forecasting due to its promising performance and capability in handling time-series data for both small- and large-scale data.     

The surface characteristics of under bump metallurgy (UBM) in electroless nickel immersion gold (ENIG) deposition

This paper discusses on the surface characteristics of each of the seven set-up process steps prior completion of under bump metallurgy (UBM) deposition. The Atomic Force Microscopy (AFM) and Scanning Electron Microscopy (SEM) techniques were used to obtain the surface roughness and morphology of each deposition process. The elemental analysis using Energy Disperse X-Ray (EDX) and precision micro-cross-section using Focus Ion Beam (FIB) was also used in this study. The first zincation process has high surface roughness but preserved surface morphology of initial thin-film surface. The second zincation provides the improved surface roughness due to the replacement aluminum layer with ion zinc in the solution. Smooth surfaces leads to better shear strength. However, second zincation will affect the Al thickness, hence excessive attacks on Al layer may cause reliability problem.     

A matheuristic for the discrete bilevel problem with multiple objectives at the lower level

In this paper, we solve a discrete bilevel problem with multiple objectives at the lower level and constraints at the upper level coupling variables of both levels. In the case of a multiobjective lower level, we deal with a set of Pareto‐efficient solutions rather than a single optimal lower level solution. To calculate the upper level objective function value, we need to select one solution out of a potentially large set of efficient lower level solutions. To avoid the enumeration of the whole set of Pareto solutions, we formulate an auxiliary mixed integer linear programming problem with a large number of constraints. We propose an iterative exact method to solve it. To find a near‐optimal upper level solution, we apply a metaheuristic. The method is tested on the discrete ()‐centroid problem with multiple objectives at the lower level.     

Biosynthesis of silver nanoparticles using Azadirachta indica leaves: characterisation and impact on Staphylococcus aureus growth and glutathione‐S‐transferase activity

Silver nanoparticles (AgNPs) are toxic to various microbes, but the mechanism of action is not fully understood. The present report explores Azadirachta indica leaf extract as a reducing agent for the rapid biosynthesis of AgNPs. The effects of AgNPs on the growth, glutathione‐S‐transferase (GST) activity, and total protein concentration in Staphylococcus aureus were investigated, as was its antibacterial activity against seven other bacterial strains. Nanoparticle synthesis was confirmed by the UV‐Vis spectrum and colour change of the solution. Scanning electron microscopy (SEM), transmission electron microscopy (TEM), dynamic light scattering (DLS), zeta potential analysis, and infrared spectroscopy were used to characterise the synthesised nanoparticles. The UV‐Visible spectrograph showed an absorbance peak at 420 nm. DLS analysis showed an average AgNP size of 159 nm and a Polydispersity Index of 0.373. SEM analysis showed spherical particle shapes, while TEM established an average AgNP size of 7.5 nm. The element analysis profile showed small peaks for calcium, potassium, zinc, chlorine, with the presence of oxygen and silver. AgNPs markedly affected the growth curves and GST activity in treated bacteria, and produced moderate antibacterial activity. Thus AgNPs synthesised from A. indica leaves can interrupt the growth curve and total protein concentration in bacterial cells.Inspec keywords: ultraviolet spectra, microorganisms, nanomedicine, visible spectra, nanoparticles, electrokinetic effects, antibacterial activity, scanning electron microscopy, infrared spectra, transmission electron microscopy, light scattering, nanofabrication, particle size, silver, enzymes, biochemistry, molecular biophysics, cellular biophysicsOther keywords: silver nanoparticles, glutathione‐S‐transferase activity, green leaves, rapid biosynthesis, total protein concentration, nanoparticle synthesis, colour change, zeta potential analysis, UV‐Visible spectrograph, DLS analysis, SEM analysis, element analysis profile, growth curve, GST activity, bacterial strains, antibacterial activity, staphylococcus aureus growth, microbes, Azadirachta azadirachta indica leaf, reducing agent, scanning electron microscopy, transmission electron microscopy, dynamic light scattering, infrared spectroscopy, absorbance peak, polydispersity index, spherical particle shapes, TEM, bacterial cells, Ag     

Enhancement of artificial neural network learning using centripetal accelerated particle swarm optimization for medical diseases diagnosis

In recent decades, artificial neural networks (ANNs) have been extensively applied in different areas such as engineering, medicine, business, education, manufacturing and so on. Nowadays, ANNs are as a hot research in medicine especially in the fields of medical disease diagnosis. To have a high efficiency in ANN, selection of an appropriate architecture and learning algorithm is very important. ANN learning is a complex task and an efficient learning algorithm has a significant role to enhance ANN performance. In this paper, a new meta-heuristic algorithm, centripetal accelerated particle swarm optimization (CAPSO), is applied to evolve the ANN learning and accuracy. The algorithm is based on an improved scheme of particle swarm algorithm and Newton’s laws of motion. The hybrid learning of CAPSO and multi-layer perceptron (MLP) network, CAPSO-MLP, is used to classify the data of nine standard medical datasets of Hepatitis, Heart Disease, Pima Indian Diabetes, Wisconsin Prognostic Breast Cancer, Parkinson’s disease, Echocardiogram, Liver Disorders, Laryngeal 1 and Acute Inflammations. The performance of CAPSO-MLP is compared with those of PSO, gravitational search algorithm and imperialist competitive algorithm on MLP. The efficiency of methods are evaluated based on mean square error, accuracy, sensitivity, specificity, area under the receiver operating characteristics curve and statistical tests of \(t\) -test and Wilcoxon’s signed ranks test. The results indicate that CAPSO-MLP provides more effective performance than the others for medical disease diagnosis especially in term of unseen data (testing data) and datasets with high missing data values.     

Effect of Treatments on the Physical and Morphological Properties of SPF/Phenolic Composites

This study aims at evaluating the physical properties and effects of fiber treatments of natural fiber reinforced polymer composite’s friction applications. Sugar palm fibers (SPFs) were used as fillers (≤ 150 µm) with phenolic resin to fabricate the composites by the hot press technique. The loading of SPFs varied from 0 to 40 vol.% with an interval of 10 vol.% in phenolic composites. The fibers were treated with sea water for 30 days, and with 0.5 M of alkaline solution for 4 hrs. Rockwell hardness, density, voids content, water/oil absorption, and moisture content were studied. Scanning electron microscopy (SEM) was used to investigate the morphology and interfacial bonding of the fiber-matrix in composites. With an increase in the SPF loading in the composites, the results indicated a decline in Rockwell hardness, an increase in water/oil absorption, and density. It was also observed that higher the density of the composites, lower was the voids content. In terms of physical properties, sea water treatment showed better improvement than alkaline treatment. The outcome of this research indicated that SPFs can be effectively used in reinforcing polymer composites, such as friction composites.     

Determination of iodine value of palm olein mixtures using differential scanning calorimetry

Differential scanning calorimetry (DSC) was used to determine the iodine value (IV) of various palm olein (PoO) mixtures. Eight different PoO mixtures, namely, PoO:PKO, PoO:CoO, PoO:PS, PoO:PO, PoO:CaO, PoO:OeO, PoO:CnO and PoO:SFO were prepared at different ratios (w/w) to give various IV (PKO represents palm kernel olein; CoO, coconut oil; PS, palm stearin; PO, palm oil; CaO, canola oil; OeO, olive oil; CnO, corn oil; and SFO, sunflower oil). Each sample was then scanned from 80 to ‐ 100 �C at ‐5 �C/min using a DSC. All the mixtures showed two exothermic peaks in their cooling thermograms, except PoO:SFO mixtures which showed three peaks. Results of stepwise multiple linear regression (SMLR) analysis showed that five independent variables extracted from each of these peaks, namely, on‐set temperature, off‐set temperature, peak temperature, peak height and peak enthalpy could predict well the IV of each mixture. The calibration models developed showed appreciable effectiveness, re‐producibility and accuracy, and specificity towards the calibration data set. A shared calibration model for each group of PoO mixtures i.e. high‐lauric (PoO:PKO and PoO:CoO), high‐palmitic (PoO:PS and PoO:PO) and high‐oleic (PoO:CaO and PoO:OeO) mixtures was also developed. SMLR analysis showed that the shared models were also capable in predicting IV of the PoO mixtures, even though the coefficient of determination, R² , was slightly lower than that of their individual models. The shared calibration models also had good reproducibility and accuracy when compared with the standard chemical method. In conclusion, DSC provides an effective method in determining IV for routine analysis in the industries, whereby one single model could be calibrated for the use of all oil and fat products that have similar chemical compositions such as high‐lauric, high‐palmitic or high‐oleic mixtures in the industry.