MultiDMet: Designing a Hybrid Multidimensional Metrics Framework to Predictive Modeling for Performance Evaluation and Feature Selection

In a competitive digital age where data volumes are increasing with time, the ability to extract meaningful knowledge from high-dimensional data using machine learning (ML) and data mining (DM) techniques and making decisions based on the extracted knowledge is becoming increasingly important in all business domains. Nevertheless, high-dimensional data remains a major challenge for classification algorithms due to its high computational cost and storage requirements. The 2016 Demographic and Health Survey of Ethiopia (EDHS 2016) used as the data source for this study which is publicly available contains several features that may not be relevant to the prediction task. In this paper, we developed a hybrid multidimensional metrics framework for predictive modeling for both model performance evaluation and feature selection to overcome the feature selection challenges and select the best model among the available models in DM and ML. The proposed hybrid metrics were used to measure the efficiency of the predictive models. Experimental results show that the decision tree algorithm is the most efficient model. The higher score of HMM (m, r) = 0.47 illustrates the overall significant model that encompasses almost all the user’s requirements, unlike the classical metrics that use a criterion to select the most appropriate model. On the other hand, the ANNs were found to be the most computationally intensive for our prediction task. Moreover, the type of data and the class size of the dataset (unbalanced data) have a significant impact on the efficiency of the model, especially on the computational cost, and the interpretability of the parameters of the model would be hampered. And the efficiency of the predictive model could be improved with other feature selection algorithms (especially hybrid metrics) considering the experts of the knowledge domain, as the understanding of the business domain has a significant impact.

Understanding the roles of Brønsted/Lewis acid sites on manganese oxide-zeolite hybrid catalysts for low-temperature NH_(3)-SCR

Although metal oxide-zeolite hybrid materials have long been known to achieve enhanced catalytic activity and selectivity in NO_(x)removal reactions through the inter-particle diffusion of intermediate species,their subsequent reaction mechanism on acid sites is still unclear and requires investigation.In this study,the distribution of Brønsted/Lewis acid sites in the hybrid materials was precisely adjusted by introducing potassium ions,which not only selectively bind to Brønsted acid sites but also potentially affect the formation and diffusion of activated NO species.Systematic in situ diffuse reflectance infrared Fourier transform spectroscopy analyses coupled with selective catalytic reduction of NO_(x)with NH_(3)(NH_(3)-SCR)reaction demonstrate that the Lewis acid sites over MnO_(x)are more active for NO reduction but have lower selectivity to N_(2)than Brønsted acids sites.Brønsted acid sites primarily produce N_(2),whereas Lewis acid sites primarily produce N_(2)O,contributing to unfavorable N_(2)selectivity.The Brønsted acid sites present in Y zeolite,which are stronger than those on MnO_(x),accelerate the NH_(3)-SCR reaction in which the nitrite/nitrate species diffused from the MnO_(x)particles rapidly convert into the N_(2).Therefore,it is important to design the catalyst so that the activated NO species formed in MnO_(x)diffuse to and are selectively decomposed on the Brønsted acid sites of H-Y zeolite rather than that of MnO_(x)particle.For the physically mixed H-MnO_(x)+H-Y sample,the abundant Brønsted/Lewis acid sites in H-MnO_(x)give rise to significant consumption of activated NO species before their inter-particle diffusion,thereby hindering the enhancement of the synergistic effects.Furthermore,we found that the intercalated K+in K-MnO_(x)has an unexpected favorable role in the NO reduction rate,probably owing to faster diffusion of the activated NO species on K-MnO_(x)than H-MnO_(x).This study will help to design promising metal oxide-zeolite hybrid catalysts by identifying the role of the acid sites in two different constituents.

Analyzing the Impact of Feature Selection on Crop Yield Prediction

In agriculture sector, machine learning has been widely used by researchers for crop yield prediction. However, it is quite difficult to identify the most critical features from a dataset. Feature selection techniques allow us to remove the extraneous and noisy features from the original feature set. The feature selection techniques help the model to focus only on the important features of the data, thus reducing execution time and improving efficiency of the model. The aim of this study is to determine relevant subset features for achieving high predictive performance by using different feature selection techniques like Filter methods, Wrapper methods and embedded methods. In this work, different feature selection techniques like Rank-based feature selection technique, weighted feature selection technique and Hybrid Feature Selection Technique have been applied to the agricultural data. The optimal feature set returned by different feature selection techniques is used for yield prediction using Linear regression, Random Forest, and Decision Tree Regressor. The accuracy of prediction obtained using the above three methods has been analyzed by using different evaluation parameters. This study helps in increasing predictive accuracy with the minimum number of features.

Spatial Heterogeneity Modeling Using Machine Learning Based on a Hybrid of Random Forest and Convolutional Neural Network (CNN)

Spatial heterogeneity refers to the variation or differences in characteristics or features across different locations or areas in space. Spatial data refers to information that explicitly or indirectly belongs to a particular geographic region or location, also known as geo-spatial data or geographic information. Focusing on spatial heterogeneity, we present a hybrid machine learning model combining two competitive algorithms: the Random Forest Regressor and CNN. The model is fine-tuned using cross validation for hyper-parameter adjustment and performance evaluation, ensuring robustness and generalization. Our approach integrates Global Moran’s I for examining global autocorrelation, and local Moran’s I for assessing local spatial autocorrelation in the residuals. To validate our approach, we implemented the hybrid model on a real-world dataset and compared its performance with that of the traditional machine learning models. Results indicate superior performance with an R-squared of 0.90, outperforming RF 0.84 and CNN 0.74. This study contributed to a detailed understanding of spatial variations in data considering the geographical information (Longitude & Latitude) present in the dataset. Our results, also assessed using the Root Mean Squared Error (RMSE), indicated that the hybrid yielded lower errors, showing a deviation of 53.65% from the RF model and 63.24% from the CNN model. Additionally, the global Moran’s I index was observed to be 0.10. This study underscores that the hybrid was able to predict correctly the house prices both in clusters and in dispersed areas.

RBF-based cluster-head selection for wireless sensor networks

The radial basis function （RBF）, a kind of neural networks algorithm, is adopted to select clusterheads. It has many advantages such as simple parallel distributed computation, distributed storage, and fast learning. Four factors related to a node becoming a cluster-head are drawn by analysis, which are energy （ energy available in each node）, number （the number of neighboring nodes）, centrality （ a value to classify the nodes based on the proximity how central the node is to the cluster）, and location （the distance between the base station and the node）. The factors are as input variables of neural networks and the output variable is suitability that is the degree of a node becoming a cluster head. A group of cluster-heads are selected according to the size of network. Then the base station broadcasts a message containing the list of cluster-heads＇ IDs to all nodes. After that, each cluster-head announces its new status to all its neighbors and sets up a new cluster. If a node around it receives the message, it registers itself to be a member of the cluster. After identifying all the members, the cluster-head manages them and carries out data aggregation in each cluster. Thus data flowing in the network decreases and energy consumption of nodes decreases accordingly. Experimental results show that, compared with other algorithms, the proposed algorithm can significantly increase the lifetime of the sensor network.

Hybrid Rice Resistant to Bacterial Leaf Blight Developed By Marker Assisted Selection

Through recurrent backcrossing in combination with molecular marker-assisted selection (MAS), restorer lines R8006 and Rl176 carrying Xa-21, a gene having broad-spectrum resistance to rice bacterial leaf blight, were selected. By crossing the two lines to CMS line Zhong 9A, two new hybrid rice combinations, Zhongyou 6 and Zhongyou 1176 were developed. The hybrids showed high resistance to diseases, good grain quality and high yielding potential in national and provincial adaptability and yield trials.

Subsurface Defect Evaluation of Selective-Laser-Melted Inconel 738LC Alloy Using Eddy Current Testing for Additive/Subtractive Hybrid Manufacturing

New materials and manufacturing technologies require applicable non-destructive techniques for quality assurance so as to achieve better performance.This study comprehensively investigated the effect of influencing factors includ-ing excitation frequency,lift-off distance,defect depth and size,residual heat,and surface roughness on the defect EC signals of an Inconel 738LC alloy produced by selective laser melting(SLM).The experimental investigations recorded the impedance amplitude and phase angle of EC signals for each defect to explore the feasibility of detecting sub-surface defects by merely analyzing these two key indicators.Overall,this study revealed preliminary qualitative and roughly quantitative relationships between influencing factors and corresponding EC signals,which provided a prac-tical reference on how to quantitively inspect subsurface defects using eddy current testing(ECT)on SLMed parts,and also made solid progress toward on-line ECT in additive/subtractive hybrid manufacturing(ASHM)for fabricating SLMed parts with enhanced quality and better performance.

ARTIFICIAL NEURAL NETWORKS BASED GEARS MATERIAL SELECTION HYBRID INTELLIGENT SYSTEM

An artificial neural networks(ANNs) based gear material selection hybrid intelligent system is established by analyzing the individual advantages and weakness of expert system (ES) and ANNs and the applications in material select of them. The system mainly consists of tow parts: ES and ANNs. By being trained with much data samples, the back propagation (BP) ANN gets the knowledge of gear materials selection, and is able to inference according to user input. The system realizes the complementing of ANNs and ES. Using this system, engineers without materials selection experience can conveniently deal with gear materials selection.

A Hybrid Feature Selection Framework for Predicting Students Performance

Student performance prediction helps the educational stakeholders to take proactive decisions and make interventions,for the improvement of quality of education and to meet the dynamic needs of society.The selection of features for student’s performance prediction not only plays significant role in increasing prediction accuracy,but also helps in building the strategic plans for the improvement of students’academic performance.There are different feature selection algorithms for predicting the performance of students,however the studies reported in the literature claim that there are different pros and cons of existing feature selection algorithms in selection of optimal features.In this paper,a hybrid feature selection framework(using feature-fusion)is designed to identify the significant features and associated features with target class,to predict the performance of students.The main goal of the proposed hybrid feature selection is not only to improve the prediction accuracy,but also to identify optimal features for building productive strategies for the improvement in students’academic performance.The key difference between proposed hybrid feature selection framework and existing hybrid feature selection framework,is two level feature fusion technique,with the utilization of cosine-based fusion.Whereas,according to the results reported in existing literature,cosine similarity is considered as the best similarity measure among existing similarity measures.The proposed hybrid feature selection is validated on four benchmark datasets with variations in number of features and number of instances.The validated results confirm that the proposed hybrid feature selection framework performs better than the existing hybrid feature selection framework,existing feature selection algorithms in terms of accuracy,f-measure,recall,and precision.Results reported in presented paper show that the proposed approach gives more than 90%accuracy on benchmark dataset that is better than the results of existing approach.

Performance of Ti6Al4V fabricated by electron beam and laser hybrid preheating and selective melting strategy

Electron beam selective melting(EBM)and selective laser melting(SLM)are regarded as significant manufacturing processes for near-net-shaped Ti6Al4V components.Generally,in the conventional EBM process,preheating is necessitated to avoid"smoke"caused by the charging of electrons.In the conventional SLM process,laser as an energy source without the risk of"smoke"can be employed to melt metal powder at low temperatures.However,because of the low absorption rate of laser,the powder bed temperature cannot reach a high level.It is difficult to obtain as-built TiAl4V with favorable comprehensive properties via conventional EBM or SLM.Hence,two types of electron beam and laser hybrid preheating(EB-LHP)combined with selective melting strategies are proposed.Using laser to preheat powder allows EBM to be performed at a low powder bed temperature(EBM-LT),whereas using an electron beam to preheat powder allows SLM to be performed at a high powder bed temperature(SLM-HT).Ti6Al4V samples are fabricated using two different manufacturing strategies(i.e.,EBM-LT and SLM-HT)and two conventional processes,i.e.,EBM at a high powder bed temperature(EBM-HT)and SLM at a low powder bed temperature(SLM-LT).The temperature-dependent surface quality,microstructure,density,and mechanical properties of the as-built Ti6Al4V samples are characterized and compared.Results show that EBM-LT Ti6Al4V exhibits a higher ultimate tensile strength(981±43 MPa)and a lower elongation(12.2%±2.3%)than EBM-HT Ti6Al4V owing to the presence ofα′martensite.The SLM-HT Ti6Al4V possesses the highest ultimate tensile strength(1,059±62 MPa)and an elongation(14.8%±4.0%)comparable to that of the EBM-HT Ti6Al4V(16.6%±1.2%).

A Hybrid Algorithm Based on PSO and GA for Feature Selection

One of the main problems of machine learning and data mining is to develop a basic model with a few features,to reduce the algorithms involved in classification’s computational complexity.In this paper,the collection of features has an essential importance in the classification process to be able minimize computational time,which decreases data size and increases the precision and effectiveness of specific machine learning activities.Due to its superiority to conventional optimization methods,several metaheuristics have been used to resolve FS issues.This is why hybrid metaheuristics help increase the search and convergence rate of the critical algorithms.A modern hybrid selection algorithm combining the two algorithms;the genetic algorithm(GA)and the Particle Swarm Optimization(PSO)to enhance search capabilities is developed in this paper.The efficacy of our proposed method is illustrated in a series of simulation phases,using the UCI learning array as a benchmark dataset.

Adaptive Agent Model with Hybrid Routing Selection Strategy for Improving the Road-Network Congestion Problem

This paper proposes an adaptive agent model with a hybrid routing selection strategy for studying the road-network congestion problem. We focus on improving those severely congested links. Firstly,a multi-agent system is built,where each agent stands for a vehicle,and it makes its routing selection by considering the shortest path and the minimum congested degree of the target link simultaneously. The agent-based model captures the nonlinear feedback between vehicle routing behaviors and road-network congestion status.Secondly,a hybrid routing selection strategy is provided,which guides the vehicle routes adapting to the realtime road-network congestion status. On this basis, we execute simulation experiments and compare the simulation results of network congestion distribution,by Floyd agent with shortest path strategy and our proposed adaptive agent with hybrid strategy. The simulation results show that our proposed model has reduced the congestion degree of those seriously congested links of road-network. Finally,we execute our model on a real road map. The results finds that those seriously congested roads have some common features such as located at the road junction or near the unique road connecting two areas. And,the results also show an effectiveness of our model on reduction of those seriously congested links in this actual road network. Such a bottom-up congestion control approach with a hybrid congestion optimization perspective will have its significance for actual traffic congestion control.

A select link analysis method based on logit–weibit hybrid model

Select link analysis provides information of where traffic comes from and goes to at selected links.This disaggregate information has wide applications in practice.The state-of-the-art planning software packages often adopt the user equilibrium(UE) model for select link analysis.However,empirical studies have repeatedly revealed that the stochastic user equilibrium model more accurately predicts observed mean and variance of choices than the UE model.This paper proposes an alternative select link analysis method by making use of the recently developed logit–weibit hybrid model,to alleviate the drawbacks of both logit and weibit models while keeping a closed-form route choice probability expression.To enhance the applicability in large-scale networks,Bell’s stochastic loading method originally developed for logit model is adapted to the hybrid model.The features of the proposed method are twofold:(1) unique O–D-specific link flow pattern and more plausible behavioral realism attributed to the hybrid route choice model and(2) applicability in large-scale networks due to the link-based stochastic loading method.An illustrative network example and a case study in a large-scale network are conducted to demonstrate the efficiency and effectiveness of the proposed select link analysis method as well as applications of O–D-specific link flow information.A visualizationmethod is also proposed to enhance the understanding of O–D-specific link flow originally in the form of a matrix.

Polyaniline/AgCl Hybrid Materials for Selective Determination of Dopamine by Electrochemical Methods

A promising electrochemical sensor based on PANI/AgCl hybrid material has been developed. The organic/inorganic hybrid material has exhibited good electrocatalytic properties by cyclic voltammetry measurement and differential pulse voltammetry. The oxidation overpotential of dopamine decreased dramatically, and the oxidation peak current increased significantly at PANI/AgCl/GCE compared to those obtained at PANI/GCE, AgCl/GCE and bare GCE, corresponding to the synergistic effect between PANI and inorganic particle AgCl. Under the optimized conditions, the linear response in the concentration range of 0.7 to 6.0 μM for the selective determination dopamine on the PANI/AgCl/GCE is obtained with a detection limit of 5.4 × 10–8 M (S/N = 3) using differential pulse voltammetry. The results indicated that the modified electrode can be used to determine dopamine without the interference from ascorbic acid and ensure high sensitivity and good selectivity.

Reconstructing proton channels via Zr-MOFs realizes highly ion-selective and proton-conductive SPEEK-based hybrid membrane for vanadium flow battery

There is an urgent need to break through the trade-off between proton conductivity and ion selectivity of proton exchange membrane(PEM)in vanadium flow battery(VFB).Proton channels in PEM are the key to controlling the ion sieving and proton conductivity in VFB.Herein,two types of proton channels are reconstructed in the hybrid membrane via introducing modified Zr-MOFs(IM-UIO-66-AS)into SPEEK matrix.Internal proton channels in IM-UIO-66-AS and interfacial proton channels between grafted imidazole groups on Zr-MOFs and SPEEK greatly improve the conductivity of the IM-UIO-66-AS/SPEEK hybrid membrane.More importantly,both reconstructed proton channels block the vanadium-ion permeation to realize enhanced ion selectivity according to the size sieving and Donnan exclusion effects,respectively.Moreover,the hybrid membrane exhibits good mechanical property and dimensional stability.Benefiting from such rational design,a VFB loading with the optimized membrane exhibits enhanced voltage efficiency of 79.9%and outstanding energy efficiency of 79.6%at 200 m A cm^(-2),and keeps a notable cycle stability for 300 cycles in the long-term cycling test.Therefore,this study provides inspiration for preparing next-generation PEMs with high ion selectivity and proton conductivity for VFB application.

Hybrid Sooty Tern Optimization and Differential Evolution for Feature Selection

In this paper,a hybrid model based on sooty tern optimization algo-rithm(STOA)is proposed to optimize the parameters of the support vector machine(SVM)and identify the best feature sets simultaneously.Feature selec-tion is an essential process of data preprocessing,and it aims to find the most rele-vant subset of features.In recent years,it has been applied in many practical domains of intelligent systems.The application of SVM in many fields has proved its effectiveness in classification tasks of various types.Its performance is mainly determined by the kernel type and its parameters.One of the most challenging process in machine learning is feature selection,intending to select effective and representative features.The main disadvantages of feature selection processes included in classical optimization algorithm are local optimal stagnation and slow convergence.Therefore,the hybrid model proposed in this paper merges the STOA and differential evolution(DE)to improve the search efficiency and con-vergence rate.A series of experiments are conducted on 12 datasets from the UCI repository to comprehensively and objectively evaluate the performance of the proposed method.The superiority of the proposed method is illustrated from dif-ferent aspects,such as the classification accuracy,convergence performance,reduced feature dimensionality,standard deviation(STD),and computation time.

Joint waveform selection and power allocation algorithm in manned/unmanned aerial vehicle hybrid swarm based on chance-constraint programming

In this paper, we propose a joint waveform selection and power allocation(JWSPA) strategy based on chance-constraint programming(CCP) for manned/unmanned aerial vehicle hybrid swarm(M/UAVHS) tracking a single target. Accordingly,the low probability of intercept(LPI) performance of system can be improved by collaboratively optimizing transmit power and waveform. For target radar cross section(RCS) prediction, we design a random RCS prediction model based on electromagnetic simulation(ES) of target. For waveform selection, we build a waveform library to adaptively manage the frequency modulation slope and pulse width of radar waveform. For power allocation,the CCP is employed to balance tracking accuracy and power resource. The Bayesian Cramér-Rao lower bound(BCRLB) is adopted as a criterion to measure target tracking accuracy. The hybrid intelli gent algorithms, in which the stochastic simulation is integrated into the genetic algorithm(GA), are used to solve the stochastic optimization problem. Simulation results demonstrate that the proposed JWSPA strategy can save more transmit power than the traditional fixed waveform scheme under the same target tracking accuracy.

Selection of wide compatible restorer lines and their application in hybrid rice breeding

Since the middle of 1980’s, wide compatibility(WC) rice lines have been screened by ricebreeders in China and applied in hybrid ricebreeding program. Several WC lines such asPecos, T984, Lunhui 422, and 02428 withideal agronomic characters were identified. Weincorporated the WC gene into restorer linesby crossing these japonica WC lines with ob-tained indica lines. Some WC restorer lineswith indica-japonica medium type were ob-tained and their application value in intersub-specific hybrid rice breeding were evaluated. 1. Effect of crossing methods on selectionefficiencies of WC restorer lines

Gametic selection in a doubled haploid population derived from anther culture of indica/japonica hybrid of Oryza Sativa L.

It is one of the key problems for application ofanther culture in hybrid breeding, geneticanalysis, and molecular mapping whether thedoubled haploid (DH) population derived fromanther culture of rice crosses represents a ran-dom array of the microspore population, i.e.whether gametic selection occurs in androgene-sis. A DH population including 132 lines de-

基于生产数据的混合流水车间动态调度方法研究

在智能制造背景下,物联网等信息技术为制造系统积累了大量数据,同时人工智能等先进方法为车间数据分析和实时控制提供了有效手段。因此,针对不相关并行机混合流水车间调度问题,提出了一种基于生产数据的动态调度方法,以实现订单完工时间最小化。首先以高质量调度方案为基础,从中提取生产特征和调度规则完成样本构建。其次使用Relief F算法过滤冗余生产特征,获得用于训练和预测的调度样本。然后采用融合鲸鱼优化算法的概率神经网络作为调度模型,实现基于调度样本的训练和预测过程。最后,实验结果表明,所提方法具有良好的特征选择能力和较高的预测精度,与其他实时调度方法相比具有更加优越的性能,可以有效地根据车间实时状态指导制造执行过程。