editorial artificial intelligence and its...

11
Editorial Artificial Intelligence and Its Applications Yudong Zhang, 1 Saeed Balochian, 2 Praveen Agarwal, 3 Vishal Bhatnagar, 4 and Orwa Jaber Housheya 5 1 School of Computer Science and Technology, Nanjing Normal University, Nanjing, Jiangsu 210023, China 2 Department of Electrical Engineering, Gonabad Branch, Islamic Azad University, Gonabad, Khorasan-e-Razavi 96916-29, Iran 3 Department of Mathematics, Anand International College of Engineering, Agra Road, Near Bassi, Jaipur, Rajasthan 303012, India 4 Ambedkar Institute of Advanced Communication Technologies and Research, Government of NCT of Delhi, Geeta Colony, Delhi 110031, India 5 Department of Chemistry, Arab American University, 19356 Jenin, Palestine Correspondence should be addressed to Yudong Zhang; [email protected] Received 18 March 2014; Accepted 18 March 2014; Published 10 April 2014 Copyright © 2014 Yudong Zhang et al. is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Artificial intelligence (AI) has revolutionized information technology. e new economy of information technology has shaped the way we are living. Recently, AI algorithms have attracted close attention of researchers and have also been applied successfully to solve problems in engineering. Nevertheless, for large and complex problems, AI algorithms consume considerable computation time due to stochastic feature of the search approaches. erefore, there is a poten- tial requirement to develop efficient algorithm to find solu- tions under the limited resources, time, and money in real- world applications. is special issue aims to report the latest advances in every aspect of artificial intelligence technology, including machine learning, data mining, computer vision, multiagent systems, evolutionary computation, and fuzzy logic. e special issue received 142 high quality submissions from different countries all over the world. All submitted papers followed the same standard (peer-reviewed by at least three independent reviewers) as applied to regular submissions to Mathematical Problems in Engineering. Due to the limited space, 57 papers were finally included. e primary guideline was to demonstrate the most significant developments on the topics of AI and to apply AI algorithms in real-life scenarios. In the paper entitled “Solving two-dimensional HP model by firefly algorithm and simplified energy function,” Y. Zhang et al. investigate traditional energy function and point out that its discrete property cannot give direction of the next step to the searching point, causing a challenge to optimization algorithms. erefore, they introduce the simplified energy function to turn traditional discrete energy function to a continuous one. e simplified energy function totals the distance between all pairs of hydrophobic amino acids. To optimize the simplified energy function, they introduce the latest swarm intelligence algorithm, the firefly algorithm (FA). e experiments take 14 sequences of different chain lengths from 18 to 100 as the dataset and compare the FA with standard genetic algorithm and immune genetic algorithm. Each algorithm runs 20 times. e averaged energy convergence results show that FA achieves the lowest values. It concludes that it is effective to solve 2D HP model by the FA and the simplified energy function. In the paper entitled “Fault diagnosis for wireless sensor by twin support vector machine,” M. Ding et al. propose a novel fault diagnosis method for wireless sensor technology by twin support vector machine (TSVM) in order to improve the diagnosis accuracy of wireless sensor. Twin SVM is a binary classifier that performs classification by using two nonparallel hyperplanes instead of the single hyperplane used in the classical SVM. However, the parameter setting in the TSVM training procedure significantly influences the classification accuracy. eir study introduces PSO as an optimization technique to simultaneously optimize the TSVM training parameter. e experimental results indicate Hindawi Publishing Corporation Mathematical Problems in Engineering Volume 2014, Article ID 840491, 10 pages http://dx.doi.org/10.1155/2014/840491

Upload: others

Post on 19-Jun-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

EditorialArtificial Intelligence and Its Applications

Yudong Zhang,1 Saeed Balochian,2 Praveen Agarwal,3

Vishal Bhatnagar,4 and Orwa Jaber Housheya5

1 School of Computer Science and Technology, Nanjing Normal University, Nanjing, Jiangsu 210023, China2Department of Electrical Engineering, Gonabad Branch, Islamic Azad University, Gonabad, Khorasan-e-Razavi 96916-29, Iran3Department of Mathematics, Anand International College of Engineering, Agra Road, Near Bassi, Jaipur, Rajasthan 303012, India4Ambedkar Institute of Advanced Communication Technologies and Research, Government of NCT of Delhi,Geeta Colony, Delhi 110031, India

5 Department of Chemistry, Arab American University, 19356 Jenin, Palestine

Correspondence should be addressed to Yudong Zhang; [email protected]

Received 18 March 2014; Accepted 18 March 2014; Published 10 April 2014

Copyright © 2014 Yudong Zhang et al. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Artificial intelligence (AI) has revolutionized informationtechnology. The new economy of information technologyhas shaped the way we are living. Recently, AI algorithmshave attracted close attention of researchers and have alsobeen applied successfully to solve problems in engineering.Nevertheless, for large and complex problems, AI algorithmsconsume considerable computation time due to stochasticfeature of the search approaches. Therefore, there is a poten-tial requirement to develop efficient algorithm to find solu-tions under the limited resources, time, and money in real-world applications.This special issue aims to report the latestadvances in every aspect of artificial intelligence technology,including machine learning, data mining, computer vision,multiagent systems, evolutionary computation, and fuzzylogic.

The special issue received 142 high quality submissionsfrom different countries all over the world. All submittedpapers followed the same standard (peer-reviewed by atleast three independent reviewers) as applied to regularsubmissions to Mathematical Problems in Engineering. Dueto the limited space, 57 papers were finally included. Theprimary guideline was to demonstrate the most significantdevelopments on the topics of AI and to apply AI algorithmsin real-life scenarios.

In the paper entitled “Solving two-dimensional HP modelby firefly algorithm and simplified energy function,” Y. Zhanget al. investigate traditional energy function and point out

that its discrete property cannot give direction of the next stepto the searching point, causing a challenge to optimizationalgorithms. Therefore, they introduce the simplified energyfunction to turn traditional discrete energy function to acontinuous one. The simplified energy function totals thedistance between all pairs of hydrophobic amino acids. Tooptimize the simplified energy function, they introduce thelatest swarm intelligence algorithm, the firefly algorithm(FA). The experiments take 14 sequences of different chainlengths from 18 to 100 as the dataset and compare theFA with standard genetic algorithm and immune geneticalgorithm. Each algorithm runs 20 times. The averagedenergy convergence results show that FA achieves the lowestvalues. It concludes that it is effective to solve 2D HP modelby the FA and the simplified energy function.

In the paper entitled “Fault diagnosis for wireless sensorby twin support vector machine,” M. Ding et al. propose anovel fault diagnosis method for wireless sensor technologyby twin support vector machine (TSVM) in order to improvethe diagnosis accuracy of wireless sensor. Twin SVM is abinary classifier that performs classification by using twononparallel hyperplanes instead of the single hyperplaneused in the classical SVM. However, the parameter settingin the TSVM training procedure significantly influencesthe classification accuracy. Their study introduces PSO asan optimization technique to simultaneously optimize theTSVM training parameter. The experimental results indicate

Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2014, Article ID 840491, 10 pageshttp://dx.doi.org/10.1155/2014/840491

Page 2: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

2 Mathematical Problems in Engineering

that the diagnosis results for wireless sensor of twin supportvector machine are better than those of SVM, ANN.

In the paper “Design of special impacting filter for mul-ticarrier ABPSK system,” Z. Chen and L. Wu propose aniterative scheme to derive the parameters of the cascade notchfilter.The cost function is determined by the cascading notchfilter’s influence on impacting filters, converting the costfunction’s least square problem to a filter parameters’ standardquadratic programming problem. Finally, a cascading notchSIF (CNSIF) designed to demodulate the ABPSK signals isrealized.

In the paper “Weight-selected attribute bagging for creditscoring,” J. Li et al. propose an improved attribute baggingmethod, weight-selected attribute bagging (WSAB), to eval-uate credit risk. Weights of attributes are first computedusing attribute evaluation method such as linear supportvector machine (LSVM) and principal component analysis(PCA). Subsets of attributes are then constructed accordingto weights of attributes. For each of attribute subsets, thelarger the weights of the attributes, the larger the proba-bilities by which they are selected into the attribute subset.Next, training samples and test samples are projected ontoeach attribute subset, respectively. A scoring model is thenconstructed based on each set of newly produced trainingsamples. Finally, all scoring models are used to vote fortest instances. An individual model that only uses selectedattributes will be more accurate because of elimination ofsome of redundant and uninformative attributes. Besides, theway of selecting attributes by probability can also guaranteethe diversity of scoring models. Experimental results basedon two credit benchmark databases show that the proposedmethod, WSAB, is outstanding in both prediction accuracyand stability, as compared to analogous methods.

In the paper entitled “A novel method for surface defectdetection of photovoltaic module based on independent com-ponent analysis,” the research of X. Zhang et al. proposesa new method for surface defect detection of photovoltaicmodule based on independent component analysis (ICA)reconstruction algorithm. Firstly, a faultless image is used asthe training image. The demixing matrix and correspondingICs are obtained by applying the ICA in the training image.Then they reorder the ICs according to the range values andreform the demixing matrix. The reformed demixing matrixis used to reconstruct the defect image. The resulting imagecan remove the background structures and enhance thelocal anomalies. Experimental results show that the proposedmethod can effectively detect the presence of defects inperiodically patterned surfaces.

In the paper entitled “Neural model with particle swarmoptimization Kalman learning for forecasting in smart grids,”A. Y. Alanis et al. discuss a novel training algorithm for a neu-ral network architecture applied to time series predictionwithsmart grids applications. The proposed training algorithm isbased on an extended Kalman filter (EKF) improved usingparticle swarm optimization (PSO) to compute the designparameters. The EKF-PSO-based algorithm is employed toupdate the synaptic weights of the neural network. The sizeof the regression vector is determined by means of theCao methodology. The proposed structure captures more

efficiently the complex nature of the wind speed, energygeneration, and electrical load demand time series thatare constantly monitored in a smart grid benchmark. Theproposed model is trained and tested using real data valuesin order to show the applicability of the proposed scheme.

In the paper entitled “A wavelet-based robust relevancevector machine based on sensor data scheduling control formodelingmine gas gushing forecasting on virtual environment,”W. Ting et al. present a wavelet-based robust relevance vectormachine based on sensor data scheduling control for mod-eling mine gas gushing forecasting. Morlet wavelet functioncan be used as the kernel function of robust relevance vectormachine. Mean percentage error has been used to measurethe performance of the proposedmethod in this study. As themean prediction error of mine gas gushing of the WRRVMmodel is less than 1.5% and themean prediction error ofminegas gushing of the RVM model is more than 2.5%, it can beseen that the prediction accuracy for mine gas gushing of theWRRVMmodel is better than that of the RVMmodel.

In the paper entitled “Practical speech emotion recognitionbased on online learning: from acted data to elicited data,” C.Huang et al. study the cross-database speech emotion recog-nition based on online learning. How to apply a classifiertrained on acted data to naturalistic data, such as eliciteddata, remains a major challenge in today’s speech emotionrecognition system. They introduce three types of differentdata sources: first, a basic speech emotion dataset collectedfrom acted speech by professional actors and actresses;second, a speaker-independent data set which contains alarge number of speakers; third, an elicited speech dataset collected from a cognitive task. Acoustic features areextracted from emotional utterances and evaluated by usingmaximal information coefficient (MIC). A baseline valenceand arousal classifier is designed based on Gaussian mixturemodels. Online training module is implemented by usingAdaBoost. While the offline recognizer is trained on theacted data, the online testing data includes the speaker-independent data and the elicited data. Experimental resultsshow that by introducing the online learning module theirspeech emotion recognition system can be better adaptedto new data, which is an important character in real worldapplications.

In the paper entitled “Visual object tracking based on2DPCA and ML,” M.-X. Jiang et al. present a novel visualobject tracking algorithm based on two-dimensional princi-pal component analysis (2DPCA) and maximum likelihoodestimation (MLE). Firstly, they introduce regularization intothe 2DPCAreconstruction anddevelop an iterative algorithmto represent an object by 2DPCA bases. Secondly, the modelof sparsity constrained MLE is established. Abnormal pixelsin the samples will be assigned with lowweights to reduce theeffects on the tracking algorithm. The object tracking resultsare obtained by using Bayesianmaximum a posteriori (MAP)probability estimation. Finally, to further reduce trackingdrift, they employ a template update strategy that combinesincremental subspace learning and the error matrix. Theirstrategy adapts the template to the appearance change of thetarget and reduces the influence of the occluded target tem-plate as well. Compared with other popular methods, their

Page 3: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

Mathematical Problems in Engineering 3

method reduces the computational complexity and is veryrobust to abnormal changes. Both qualitative and quantitativeevaluations on challenging image sequences demonstrate thatthe proposed tracking algorithm achieves more favorableperformance than several state-of-the-art methods.

In the paper entitled “Artificial bee colony algorithmmerged with pheromone communication mechanism for the 0-1 multidimensional knapsack problem,” J. Ji et al. propose anew artificial bee colony (ABC) algorithm for the 0-1 mul-tidimensional knapsack problem (MKP 01). The new ABCalgorithm introduces a novel communication mechanismamong bees, which bases on the updating and diffusion ofinductive pheromone produced by bees. In a number ofexperiments and comparisons, their approach obtains betterquality solutions in shorter time than the ABC algorithmwithout the mechanism.They also compare the solution per-formance of the proposed approach against some stochasticapproaches recently reported in the literature. Computationalresults demonstrate the superiority of the newABC approachover all the other approaches.

In another paper, Q. Zhu and L.-N. Wu present“Weighted-bit-flipping-based sequential scheduling decodingalgorithms for LDPC codes.”Their paper proposes a sequentialscheduling algorithm based on weighted bit-flipping (WBF)algorithm for the sake of improving the convergence speed.Notoriously, WBF is a low-complexity and simple algorithm.They combine it with belief propagation (BP) to obtainadvantages of these two algorithms. Flipping function usedin WBF is borrowed to determine the priority of scheduling.Simulation results show that it can provide a good tradeoffbetween FER performance and computation complexity forshort-length low-density parity-check (LDPC) codes.

In another paper, the research of A. Laudani et al.entitled “CFSO3: a new supervised swarm-based optimizationalgorithm,” presents CFSO3, an optimization heuristic withinthe class of the swarm intelligence, based on a synergyamong three different features of the Continuous flock-of-starlings optimization. One of the main novelties is that thisoptimizer is no more a classical numerical algorithm sinceit now can be seen as a continuous dynamic system, whichcan be treated by using all the mathematical instrumentsavailable for managing state equations. In addition, CFSO3allows passing from stochastic approaches to superviseddeterministic ones since the random updating of parameters,a typical feature for numerical swarm-based optimizationalgorithms, is now fully substituted by a supervised strategy:in CFSO3 the tuning of parameters is a priori designedfor obtaining both exploration and exploitation. Indeed theexploration, that is, the escaping from a local minimum, aswell as the convergence and the refinement to a solution canbe designed simply bymanaging the eigenvalues of the CFSOstate equations. Virtually in CFSO3, just the initial values ofpositions and velocities of the swarm members have to berandomly assigned. Both standard and parallel versions ofCFSO3 together with validations on classical benchmarks arepresented.

In the paper “UCAV path planning by fitness-scalingadaptive chaotic particle swarm optimization,” Y. Zhang et

al. propose a fitness-scaling adaptive chaotic particle swarmoptimization (FAC-PSO) approach as a fast and robustapproach for the task of path planning of UCAVs. TheFAC-PSO employed the fitness-scaling method, the adaptiveparameter mechanism, and the chaotic theory. Experimentsshow that the FAC-PSO is more robust and costs lesstime than elite genetic algorithm with migration, simulatedannealing, and chaotic artificial bee colony. Moreover, theFAC-PSO performs well on the application of dynamic pathplanning when the threats cruise randomly and on theapplication of 3D path planning.

In the paper entitled “Semisupervised clustering fornetworks based on fast affinity propagation,” M. Zhu et al.propose a semisupervised clustering algorithm for networksbased on fast affinity propagation (SCAN-FAP), which isessentially a kind of similarity metric learning method.Firstly, they define a new constraint similarity measureintegrating the structural information and the pairwiseconstraints, which reflects the effective similarities betweennodes in networks. Then, taking the constraint similaritiesas input, they propose a fast affinity propagation algorithmthat keeps the advantages of the original affinity propagationalgorithmwhile increasing the time efficiency by passing onlythe messages between certain nodes. Finally, by extensiveexperimental studies, they demonstrate that the proposedalgorithm can take full advantage of the prior knowledge andimprove the clustering quality significantly. Furthermore, theproposed algorithm has a superior performance to some ofthe state-of-the-art approaches.

In the paper entitled “Research on the production schedul-ing optimization for virtual enterprises,” M. Huang et al.establish a partner selection model based on an improvedant colony algorithm, then present a production schedulingframework with two layers as global scheduling and localscheduling for virtual enterprise, and give a global schedulingmathematical model with the smallest total production timebased on it. An improved genetic algorithm is proposed inthe model to solve the time complexity of virtual enterpriseproduction scheduling. The experimental results validate theoptimization of themodel and the efficiency of the algorithm.

In the paper entitled “Surface defect target identificationon copper strip based on adaptive genetic algorithm and featuresaliency,” X. Zhang et al. propose a new surface defect targetidentification method for copper strip based on adaptivegenetic algorithm (AGA) and feature saliency. First, thestudy uses gray level cooccurrence matrix (GLCM) and HUinvariant moments for feature extraction. Then, adaptivegenetic algorithm, which is used for feature selection, isevaluated and discussed. In AGA, total error rates and falsealarm rates are integrated to calculate the fitness value,and the probability of crossover and mutation is adjusteddynamically according to the fitness value.At last, the selectedfeatures are optimized in accordance with feature saliencyand are inputted into a support vector machine (SVM).Furthermore, for comparison, they conduct experimentsusing the selected optimal feature subsequence (OFS) andthe total feature sequence (TFS) separately.The experimentalresults demonstrate that the proposed method can guarantee

Page 4: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

4 Mathematical Problems in Engineering

the correct rates of classification and can lower the false alarmrates.

In the paper entitled “Interesting activities discovery formoving objects based on collaborative filtering,” G. Yuan et al.propose a method of interesting activities discovery basedon collaborative filtering. First, the interesting degree ofthe objects’ activities is calculated comprehensively. Then,combined with the newly proposed hybrid collaborativefiltering, similar objects can be computed and all kinds ofinteresting activities can be discovered. Finally, potentialactivities are recommended according to their similar objects.The experimental results show that the method is effectiveand efficient in finding objects’ interesting activities.

In the paper entitled “Optimal design of signal controlledroad networks using differential evolution optimization algo-rithm,” H. Ceylan proposes a traffic congestion minimizationmodel in which the traffic signal setting optimization is per-formed through a combined simulation-optimizationmodel.In this model, the TRANSYT traffic simulation softwareis combined with differential evolution (DE) optimizationalgorithm, which is based on the natural selection paradigm.In this context, the equilibrium network design (EQND)problem is formulated as a bilevel programming problemin which the upper level is the minimization of the totalnetwork performance index. In the lower level, the traf-fic assignment problem, which represents the route choicebehavior of the road users, is solved using the path flowestimator (PFE) as a stochastic user equilibrium assessment.The solution of the bilevel EQND problem is carried out bythe proposed differential evolution and TRANSYT with PFE,the so-called DETRANSPFE model, on a well-known signalcontrolled test network. Performance of the proposed modelis compared to that of two previous works where the EQNDproblem has been solved by genetic-algorithms- (GAs-) andharmony-search- (HS-) based models. Results show that theDETRANSPFE model outperforms the GA- and HS-basedmodels in terms of the network performance index and thecomputational time required.

In the paper entitled “A wavelet kernel-based primal twinsupport vector machine for economic development prediction,”F. Su and H. Shang propose an economic developmentprediction method based on the wavelet kernel-based primaltwin support vector machine algorithm. As gross domes-tic product (GDP) is an important indicator to measureeconomic development, economic development predictionmeans GDP prediction in this study. The wavelet kernel-based primal twin support vector machine algorithm cansolve two smaller sized quadratic programming problemsinstead of solving a large one as in the traditional supportvector machine algorithm. Economic development data ofAnhui province from 1992 to 2009 are used to study theprediction performance of the wavelet kernel-based primaltwin support vector machine algorithm. The comparison ofmean error of economic development prediction betweenwavelet kernel-based primal twin support vector machineand traditional support vector machine models trained bythe training samples with the 3–5-dimensional input vectors,respectively, is given in this paper. The testing results showthat the economic development prediction accuracy of the

wavelet kernel-based primal twin support vector machinemodel is better than that of traditional support vectormachine.

In the paper entitled “Multiagent reinforcement learningwith regret matching for robot soccer,” Q. Liu et al. pro-pose a novel multiagent reinforcement learning (MARL)algorithm Nash-Q learning with regret matching, in whichregret matching is used to speed up the well-known MARLalgorithm Nash-Q learning. It is of importance to choose thesuitable strategy of action selection to balance explorationand exploitation, with the aim of enhancing the ability ofonline learning of Nash-Q learning. In Markov game thejoint action of agents adopting regret matching algorithmcan converge to a group of points of no-regret that canbe viewed as coarse correlated equilibrium which includesNash equilibrium in essence. It is can be inferred that regretmatching can guide exploration of the state-action space sothat the rate of convergence of Nash-Q learning algorithmcan be increased. Simulation results on robot soccer validatethat compared to original Nash-Q learning algorithm, theuse of regret matching during the learning phase of Nash-Qlearning has excellent ability of online learning and results insignificant performance in terms of scores, average reward,and policy convergence.

In another paper, X. Hu et al. present “Emotion expressionof robot with personality.” They build a robot emotionalexpressionmodel based on hiddenMarkovmodel (HMM) toenable robots that have different personalities to respond in amore satisfactory emotional level. Gross emotion regulationtheory and five factors model (FFM) that are the theoreticalbasis are firstly described. The importance of the personalityeffect on the emotion expression process is proposed, andhow to make the effect quantization is discussed. After that,the algorithmofHMMis used to describe the process of emo-tional state transition and expression, and the performancetransferring probability affected by personality is calculated.At last, the algorithm model is simulated and applied ina robot platform. The results prove that the emotionalexpression model can acquire human-like expressions andimprove the human-computer interaction.

In another paper, the research of Y.-H. Kim and Y. Yoonentitled “Context prediction of mobile users based on time-inferred pattern networks: a probabilistic approach” presentsa probabilistic method of predicting context of mobile usersbased on their historic context data. The presented methodpredicts general context based on probability theory througha novel graphical data structure, which is a kind of weighteddirected multigraphs. User context data are transformed intothe new graphical structure, in which each node representsa context or a combined context and each directed edgeindicates a context transfer with the time weight inferredfrom corresponding time data. They also consider the peri-odic property of context data and devise a good solution tocontext data with such property. Test shows the merits of thepresented method.

In the paper “Tundish cover flux thickness measurementmethod and instrumentation based on computer vision incontinuous casting tundish,” M. Lu et al. specifically designand build instrumentation and present a novel method to

Page 5: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

Mathematical Problems in Engineering 5

measure the tundish cover flux (TCF) thickness. The instru-mentation is composed of a measurement bar, a mechanicaldevice, a high-definition industrial camera, a Siemens S7-200 programmable logic controller (PLC), and a computer.Their measurement method is based on the computer visionalgorithms, including image denoising method, monocularrange measurement method, scale invariant feature trans-form (SIFT), and image gray gradient detection method.Using the present instrumentation and method, images inthe continuous casting (CC) tundish can be collected bycamera and transferred to computer to do image processing.Experiments show that the proposed instrumentation andmethod worked well at scene of steel plants, can accuratelymeasure the thickness of TCF, and overcome the disadvan-tages of traditional measurement methods, or even replacethe traditional ones.

In the paper entitled “Ripple-spreading network modeloptimization by genetic algorithm,” X.-B. Hu et al. attempt toapply genetic algorithm (GA) to tune the values of ripple-spreading related parameters (RSRPs), so that the ripple-spreading network model (RSNM) may generate these twomost important network topologies. The study demonstratesthat, once RSRPs are properly tuned by GA, the RSNM iscapable of generating both network topologies and thereforehas a great flexibility to study many real-world complexnetwork systems.

In the paper entitled “A hybrid bat algorithm with pathrelinking for capacitated vehicle routing problem,” Y. Zhouet al. propose a hybrid bat algorithm with path relinking(HBA-PR) to solve capacitated vehicle routing problem(CVRP).TheHBA-PR is constructed based on the frameworkof continuous bat algorithm; the greedy randomized adaptivesearch procedure (GRASP) and path relinking are effectivelyintegrated into bat algorithm. Moreover, in order to furtherimprove the performance, the random subsequences andsingle-point local search are operated with certain loudness(probability). In order to verity the validity of their method,several classical CVRP instances from three classes of bench-marks are selected to test its efficiency compared with otherexisting methods. Experimental results and comparisonsshow that the HBA-PR is effective for CVRP.

In the paper entitled “Study on semi-parametric statisticalmodel of safety monitoring of cracks in concrete dams,” C.Gu et al. consider that cracks are one of the hidden dangersin concrete dams. The study on safety monitoring modelsof concrete dam cracks has always been difficult. Using theparametric statistical model of safety monitoring of cracksin concrete dams, with the help of the semiparametricstatistical theory, and considering the abnormal behaviors ofthese cracks, the semiparametric statistical model of safetymonitoring of concrete dam cracks is established to overcomethe limitation of the parametric model in expressing theobjective model. Previous projects show that the semipara-metric statistical model has a stronger fitting effect and has abetter explanation for cracks in concrete dams than the para-metric statistical model. However, when used for forecast,the forecast capability of the semiparametric statistical modelis equivalent to that of the parametric statistical model. Themodeling of the semiparametric statistical model is simple,

has a reasonable principle, and has a strong practicality, witha good application prospect in the actual project.

In the paper entitled “Efficient secure multiparty compu-tation protocol for sequencing problem over insecure channel,”Y. Sun et al. believe that secure multiparty computation ismore and more popular in electronic bidding, anonymousvoting, and online auction, as a powerful tool in solvingprivacy preserving cooperative problems. Privacy preservingsequencing problem that is an essential link is regardedas the core issue in these applications. However, due tothe difficulties of solving multiparty privacy preservingsequencing problem, related secure protocol is extremelyrare. In order to break this deadlock, their paper presentsan efficient secure multiparty computation protocol for thegeneral privacy-preserving sequencing problem based onsymmetric homomorphic encryption. The result is of valuenot only in theory, but also in practice.

In the paper entitled “Nighttime fire/smoke detectionsystem based on a support vector machine,” C.-C. Ho statesthat a laser light can be projected into the monitored field ofview and the returning projected light section image can beanalyzed to detect fire and/or smoke, in order to overcomethe nighttime limitations of video smoke detection methods.If smoke appears within the monitoring zone created fromthe diffusion or scattering of light in the projected path, thecamera sensor receives a corresponding signal.The successiveprocessing steps of the proposed real-time algorithm usethe spectral, diffusing, and scattering characteristics of thesmoke-filled regions in the image sequences to register theposition of possible smoke in a video. Characterization ofsmoke is carried out by a nonlinear classification methodusing a support vector machine, and this is applied to iden-tify the potential fire/smoke location. Experimental resultsin a variety of nighttime conditions demonstrate that theproposed fire/smoke detection method can successfully andreliably detect fires by identifying the location of smoke.

In the paper entitled “Improved SpikeProp for usingparticle swarm optimization,” F. Y. H. Ahmed et al. derivea novel supervised learning rule for SpikeProp to overcomethe discontinuities introduced by the spiking thresholding.Their algorithm is based on an error-backpropagation learn-ing rule suited for supervised learning of spiking neuronsthat use exact spike time coding. The SpikeProp is able todemonstrate the spiking neurons that can perform complexnonlinear classification in fast temporal coding. Their studyproposes enhancements of SpikeProp learning algorithm forsupervised training of spiking networks which can deal withcomplex patterns. The proposed methods include the Spike-Prop particle swarm optimization (PSO) and angle drivendependency learning rate. These methods are presented toSpikeProp network for multilayer learning enhancement andweights optimization. Input and output patterns are encodedas spike trains of precisely timed spikes, and the networklearns to transform the input trains into target output trains.With these enhancements, the proposed methods outper-formed other conventional neural network architectures.

In the paper entitled “LGMS-FOA: an improved fruitfly optimization algorithm for solving optimization problems,”D. Shan et al. empirically study the performance of fruit

Page 6: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

6 Mathematical Problems in Engineering

fly optimization algorithm (FOA). Six different nonlinearfunctions are selected as testing functions. The experimentalresults illustrate that FOA cannot solve complex optimizationproblems effectively. In order to enhance the performance ofFOA, an improved FOA (named LGMS-FOA) is proposed.Simulation results and comparisons of LGMS-FOAwith FOAand other metaheuristics show that LGMS-FOA can greatlyenhance the searching efficiency and greatly improve thesearching quality.

In another paper, Y.Wang et al. present “Robust quadraticregression and its application to energy-growth consumptionproblem.” The paper proposes a robust quadratic regressionmodel to handle the statistics inaccuracy. Unlike the tra-ditional robust statistic approaches that mainly focus oneliminating the effect of outliers, the proposedmodel employsthe recently developed robust optimizationmethodology andtries to minimize the worst-case residual errors. First, theygive a solvable equivalent semidefinite programming for therobust least square model with ball uncertainty set. Thenthe result is generalized to robust models under 𝑙

1- and

𝑙∞-norm criteria with general ellipsoid uncertainty sets. In

addition, they establish a robust regression model for percapita GDP and energy consumption in the energy-growthproblem under the conservation hypothesis. Finally, numer-ical experiments are carried out to verify the effectivenessof the proposed models and demonstrate the effect of theuncertainty perturbation on the robust models.

In another paper, the research of L.Wang entitled “Extrac-tion of belief knowledge from a relational database for quanti-tative Bayesian network inference,” proposes the integrationof a Bayesian network (BN) with the functional dependency(FD) discovery technique based on a three-phase learningframework. Association rule analysis is employed to discoverFDs and expert knowledge encoded within a BN; that is, keyrelationships between attributes are emphasized. Moreover,the BN can be updated by using an expert-driven annotationprocess wherein redundant nodes and edges are removed.Experimental results show the effectiveness and efficiency ofthe proposed approach.

In the paper “Identification of code-switched sentences andwords using language modeling approaches,” L.-C. Yu et al.propose a language modeling approach to the problem ofcode-switching language processing, dividing the probleminto two subtasks: the detection of code-switched sentencesand the identification of code-switched words in sentences.A code-switched sentence is detected on the basis of whetherit contains words or phrases from another language. Oncethe code-switched sentences are identified, the positions ofthe code-switched words in the sentences are then identi-fied. Experimental results show that the language modelingapproach achieved an F-measure of 80.43% and an accuracyof 79.01% for detecting Mandarin-Taiwanese code-switchedsentences. For the identification of code-switched words, theword-based and POS-based models achieved F-measures of41.09% and 53.08%, respectively.

In the paper entitled “Knowledge mining based on envi-ronmental simulation applied to wind farm power forecasting,”D. Niu et al. propose a self-organizing map (SOM) combinedwith rough set theory clustering technique (RST) to extract

the relative knowledge and to choose themost similar historysituation and efficient data for wind power forecasting withnumerical weather prediction (NWP). Through integratingthe SOM and RST methods to cluster the historical datainto several classes, the approach could find the similardays and excavate the hidden rules. According to the datareprocessing, the selected samples will improve the forecastaccuracy echo state network (ESN) trained by the class of theforecasting day that is adopted to forecast the wind poweroutput accordingly. The developed methods are applied to acase of power forecasting in a wind farm located in northwestof China with wind power data from April 1, 2008, to May 6,2009. In order to verify its effectiveness, the performance ofthe proposed method is compared with the traditional back-propagation neural network (BP). The results demonstratedthat knowledge mining led to a promising improvement inthe performance for wind farm power forecasting.

In the paper entitled “Matching cost filtering for densestereo correspondence,” Y. Lin et al. propose a new cost-aggregation module to compute the matching responses forall the image pixels at a set of sampling points generatedby a hierarchical clustering algorithm. The complexity ofthis implementation is linear both in the number of imagepixels and in the number of clusters. Experimental resultsdemonstrate that the proposed algorithm outperforms state-of-the-art localmethods in terms of both accuracy and speed.Moreover, performance tests indicate that parameters such asthe height of the hierarchical binary tree and the spatial andrange standard deviations have a significant influence on timeconsumption and the accuracy of disparity maps.

In the paper entitled “Amultilayer hiddenMarkovmodels-based method for human-robot interaction,” C. Tao and G.Liu propose a continuous gesture recognition approachbased on multilayer hidden Markov models (MHMMs) toachieve human-robot interaction (HRI) by using gestures.The method consists of two parts. One part is gesturespotting and segment module, the other part is continuousgesture recognition module. Firstly, a Kinect sensor is usedto capture 3D acceleration and 3D angular velocity data ofhand gestures. Then, feed-forward neural networks (FNNs)and a threshold criterion are used for gesture spotting andsegment, respectively. Afterwards, the segmented gesturesignals are, respectively, preprocessed and vector symbolizedby a sliding window and a K-means clustering method.Finally, symbolized data are sent into lower hidden Markovmodels (LHMMs) to identify individual gestures, and, then,a Bayesian filter with sequential constraints among gesturesin upper hidden Markov models (UHMMs) is used tocorrect recognition errors created in LHMMs. Five prede-fined gestures are used to interact with a Kinect mobilerobot in experiments. The experimental results show thatthe proposed method not only has good effectiveness andaccuracy, but also has favorable real-time performance.

In the paper entitled “Nonlinear predictive control ofmass moment aerospace vehicles based on ant colony geneticalgorithm optimization,” X. Zhang et al. propose a novel kindofNPCparameters optimization strategy based on ant colonygenetic algorithm (ACGA), aiming at the parameters of NPCthat is generally used the trial-and-error method to optimize

Page 7: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

Mathematical Problems in Engineering 7

and design. The method for setting NPC parameters withACA in which the routes of ants are optimized by the geneticalgorithm (GA) is derived. And then, a detailed realizedprocess of this method is also presented. Furthermore, thisoptimization algorithm of the NPC parameters is applied tothe flight control system of MMAV. The simulation resultsshow that the system not only meets the demands of time-response specifications but also has excellent robustness.

In the paper entitled “Modeling and analysis of the weldbead geometry in submerged arc welding by using adaptiveneurofuzzy inference system,” N. Akkas et al. aim at obtaininga relationship between the values defining bead geometry andthe welding parameters and also to select optimum weldingparameters. For this reason, an experimental study has beenrealized. The welding parameters such as the arc current,arc voltage, and welding speed that have the most effecton bead geometry are considered, and the other parametersare held as constant. Four, three, and five different valuesfor the arc current, the arc voltage, and welding speed areused, respectively. Therefore, sixty samples made of St 52-3material are prepared. The bead geometries of the samplesare analyzed, and the thickness and penetration values ofthe weld bead are measured. Then, the relationship betweenthe welding parameters is modeled by using artificial neuralnetwork (ANN) and neurofuzzy system approach. Eachmodel is checked for its adequacy by using test data thatare selected from experimental results. Then, the modelsdeveloped are compared with regard to accuracy. In addition,the appropriate welding parameters values can be easilyselected when the models improve.

In the paper entitled “Genetic pattern search and its appli-cation to brain image classification,” Y. Zhang et al. propose anovel global optimization method based on the combinationof genetic algorithm (GA) and generalized pattern search(PS) algorithm, to find globalminimal pointsmore effectivelyand rapidly. The idea lies in the facts that GA tends to bequite good at finding generally good global solutions butquite inefficient in finding the last few mutations for theabsolute optimum and that PS is quite efficient in findingabsolute optimum in a limited region. The novel algorithm,named as genetic pattern search (GPS), employs the GA asthe search method at every step of PS. Experiments on fivedifferent classical benchmark functions (consisting of Hump,Powell, Rosenbrock, Schaffer, and Woods) demonstrate thatthe proposed GPS is superior to improved GA and improvedPS with respect to success rate. They apply the GPS to theclassification of normal and abnormal structural brain MRIimages. The results indicate that GPS exceeds BP, MBP, IGA,and IPS in terms of classification accuracy. This suggests thatGPS is an effective and viable global optimizationmethod andcan be applied to brain MRI classification.

In the paper entitled “Solving a novel inventory locationmodel with stochastic constraints and (R, s, S) inventorycontrol policy,” G. Cabrera et al. solve a novel inventory-locationmodel with a stochastic capacity constraint based ona periodic inventory control (ILM-PR) policy. The ILM-PRpolicy implies several changes with regard to other previousmodels proposed in the literature, which consider continuousreview as their inventory policy. One of these changes is

the inclusion of the undershoot concept, which has notbeen considered in previous ILM models in the literature.Based on the model, they design a distribution networkfor a two-level supply chain, addressing both warehouselocation and customer assignment decisions, whilst takinginto consideration several aspects of inventory planning, inparticular, evaluating the impact of the inventory controlreview period on the network configuration and systemcosts. Because the model is a very hard-to solve combina-torial nonlinear optimization problem, they implementedtwo heuristics to solve it, namely, Tabu Search and ParticleSwarm Optimization.These approaches are tested over smallinstances in which they are able to find the optimal solutionin just a few seconds. Because the model is a new one, a setof medium-size instances is provided that can be useful asa benchmark in future research. The heuristics show a goodconvergence rate when applied to those instances.The resultsconfirm that decision making over the inventory controlpolicy has effects on the distribution network design.

In another paper, S. Dinc et al. present “Hyperspectralimage classification using kernel Fukunaga-Koontz transform.”The paper presents a novel approach for the hyperspec-tral imagery (HSI) classification problem, using KernelFukunaga-Koontz transform (K-FKT). The Kernel basedFukunaga-Koontz transform offers higher performance forclassification problems due to its ability to solve nonlineardata distributions. K-FKT is realized in two stages: trainingand testing. In the training stage, unlike classical FKT,samples are relocated to the higher dimensional kernel spaceto obtain a transformation from nonlinear distributed datato linear form. This provides a more efficient solution tohyperspectral data classification. The second stage, testing,is accomplished by employing the Fukunaga-Koontz trans-formation operator to find out the classes of the real worldhyperspectral images. In experiment section, the improvedperformance ofHSI classification technique, K-FKT, has beentested comparing other methods such as the classical FKTand three types of support vector machines (SVMs).

In another paper, the research of A. El Mobacher et al.entitled “Entropy-based and weighted selective SIFT clusteringas an energy aware framework for supervised visual recog-nition of man-made structures,” presents uSee, a supervisedlearning framework which exploits the symmetrical andrepetitive structural patterns in buildings to identify subsetsof relevant clusters formed by these keypoints. Once asmart phone captures an image, uSee preprocesses it usingvariations in gradient angle- and entropy-based measuresbefore extracting the building signature and comparing itsrepresentative SIFT keypoints against a repository of buildingimages. Experimental results on 2 different databases confirmthe effectiveness of uSee in delivering, at a greatly reducedcomputational cost, the high matching scores for buildingrecognition that local descriptors can achieve. With only14.3% of image SIFT keypoints, uSee exceeded prior literatureresults by achieving an accuracy of 99.1% on the ZurichBuilding Database with no manual rotation, thus savingsignificantly on the computational requirements of the taskat hand.

Page 8: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

8 Mathematical Problems in Engineering

In the paper “Application of adaptive extended Kalmansmoothing on INS/WSN integration system for mobile robotindoors,” X. Chen et al. propose the inertial navigationsystems (INS)/wireless sensor network (WSN) integrationsystem of mobile robot for navigation information indoorsaccurately and continuously.The Kalman filter (KF) is widelyused for real-time applications with the aim of gainingoptimal data fusion. In order to improve the accuracy of thenavigation information, they propose an adaptive extendedKalman smoothing (AEKS) which utilizes inertial measuringunits (IMUs) and ultrasonic positioning system.The adaptiveextended Kalman filter (AEKF) is used to improve theaccuracy of forward Kalman filtering (FKF) and backwardKalman filtering (BKF), and then the AEKS and the averagefilter are used between two output timings for the onlinesmoothing. Several real indoor tests are done to assess theperformance of the proposed method. The results show thatthe proposedmethod can reduce the error comparedwith theINS-only, least squares (LS) solution, and AEKF.

In the paper entitled “The study of reinforcement learningfor traffic self-adaptive control under multiagent Markov gameenvironment,” L.-H. Xu et al. cast traffic self-adaptive controlas a multiagent Markov game problem. The design employstraffic signal control agent (TSCA) for each signalizedintersection that coordinates with neighboring TSCAs. Amathematical model for TSCAs’ interaction is built basedon nonzero-sum Markov game that has been applied to letTSCAs learn how to cooperate. A multiagent Markov gamereinforcement learning approach is constructed on the basisof single-agent Q-learning. Their method lets each TSCAlearn to update its Q-values under the joint actions andimperfect information. The convergence of the proposedalgorithm is analyzed theoretically. The simulation resultsshow that the proposed method is convergent and effectivein realistic traffic self-adaptive control setting.

In the paper entitled “Research on ISFLA-based optimalcontrol strategy for the coordinated charging of EV batteryswap station,” X. Huang et al. maintain that electric vehicles(EVs) could be a good measure against energy shortages andenvironmental pollution as an important component of thesmart grid. A main way of energy supply to EVs is to swapbattery from the swap station. Based on the characteristicsof EV battery swap station, the coordinated charging optimalcontrol strategy is investigated to smooth the load fluctuation.Shuffled frog leaping algorithm (SFLA) is an optimizationmethod inspired by the memetic evolution of a group offrogs when seeking food. An improved shuffled frog leapingalgorithm (ISFLA) with the reflecting method to deal withthe boundary constraint is proposed to obtain the solution ofthe optimal control strategy for coordinated charging. Basedon the daily load of a certain area, the numerical simulationsincluding the comparison of PSO and ISFLA are carried outand the results show that the presented ISFLA can effectivelylower the peak-valley difference and smooth the load profilewith the faster convergence rate and higher convergenceprecision.

In the paper entitled “Composite broadcasting and rangingvia a satellite dual-frequencyMPPSK system,” Y. Yao et al. pro-pose the design of dual frequency M-ray position phase shift

keying (MPPSK) system that is suitable for performing bothdata transmission and range measurement. The approach isbased on MPPSK modulation waveforms utilized in digitalvideo broadcasting. In particular, requirements that allowfor employing such signals for range measurements withhigh accuracy and high range are investigated. In addition,the relationship between the frequency difference of dualfrequency MPPSK system and range accuracy is discussed.Moreover, the selection of MPPSKmodulation parameter fordata rate and ranging is considered. In addition to theoreticalconsiderations, the paper presents system simulations andmeasurement results of new systems, demonstrating the highspectral utilization of integrated broadcasting and rangingapplications.

In the paper entitled “A genetic-algorithms-basedapproach for programming linear and quadratic optimizationproblems with uncertainty,” W. Jin et al. propose a genetic-algorithms-based approach as an all-purpose problem-solving method for operation programming problemsunder uncertainty. The proposed method is applied formanagement of a municipal solid waste treatment system.Compared to the traditional interactive binary analysis,this approach has fewer limitations and is able to reducethe complexity in solving the inexact linear programmingproblems and inexact quadratic programming problems.The implementation of this approach is performed usingthe genetic algorithm solver of MATLAB (trademark ofMathWorks). The paper explains the genetic-algorithms-based method and presents details on the computationprocedures for each type of inexact operation programmingproblems. A comparison of the results generated by theproposed method based on genetic algorithms with thoseproduced by the traditional interactive binary analysismethod is also presented.

In the paper entitled “Solving the balanced academiccurriculumproblemusing theACOmetaheuristic,” J.-M. Rubioet al. consider that the balanced academic curriculum prob-lem consists in the assignation of courses to academic periodssatisfying all the load limits and prerequisite constraints.Theypresent the design of a solution to the balanced academiccurriculum problem based on the ACO metaheuristic, inparticular via the best-worst ant system. They provide anexperimental evaluation that illustrates the effectiveness ofthe proposed approach on a set of classic benchmarks as wellas on real instances.

In the paper entitled “Hybrid functional-neural approachfor surface reconstruction,” A. Iglesias and A. Galvez intro-duce a new hybrid functional-neural approach for surfacereconstruction. The approach is based on the combinationof two powerful artificial intelligence paradigms: on onehand, they apply the popular Kohonen neural network toaddress the data parameterization problem. On the otherhand, they introduce a new functional network, calledNURBS functional network, whose topology is aimed atreproducing faithfully the functional structure of the NURBSsurfaces. These neural and functional networks are appliedin an iterative fashion for further surface refinement. Thehybridization of these two networks provides a powerfulcomputational approach to obtain a NURBS fitting surface

Page 9: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

Mathematical Problems in Engineering 9

to a set of irregularly sampled noisy data points within aprescribed error threshold. The method has been applied totwo illustrative examples. The experimental results confirmthe good performance of their approach.

In the paper entitled “An algorithm for mining of associa-tion rules for the information communication network alarmsbased on swarm intelligence,” Y. Wang et al. analyze thedata characteristics and association logic of the networkalarms. Besides, the alarm data are preprocessed and themain standardization information fields are screened. TheAPPSO algorithm is proposed on the basis of combiningthe evaluationmethod for support and confidence coefficientin the Apriori (AP) algorithm as well as the particle swarmoptimization (PSO) algorithm. By establishing a sparse linkedlist, the algorithm is able to calculate the particle support thusfurther improving the performance of the APPSO algorithm.Based on the test for the network alarm data, it is discoveredthat rational setting of the particle swarm scale and numberof iterations of the APPSO algorithm can be used to minethe vast majority and even all of the association rules and themining efficiency are significantly improved, compared withApriori algorithm.

In another paper, S.-F. Jiang et al. present “Structuralreliability assessment by integrating sensitivity analysis andsupport vector machine.” This paper proposes a structuralreliability assessmentmethod by the use of sensitivity analysis(SA) and support vector machine (SVM). The sensitivityanalysis is firstly applied to assess the effect of random vari-ables on the values of performance function, while the small-influence variables are rejected as input vectors of SVM.Then, the trained SVM is used to classify the input vectors,which are produced by sampling the residual variables basedon their distributions. Finally, the reliability assessment isimplemented with the aid of reliability theory. A 10-barplanar truss is used to validate the feasibility and efficiencyof the proposed method, and a performance comparison ismade with other existing methods. The results show thatthe proposed method can largely save the runtime with lessreduction of the accuracy; furthermore, the accuracy usingthe proposed method is the highest among the methodsemployed.

In a paper, the research of C. Liu et al. entitled “Incre-mental tensor principal component analysis for handwrittendigit recognition” proposes incremental tensor principal com-ponent analysis (ITPCA) based on updated-SVD techniquealgorithm. This paper proves the relationship between PCA,2DPCA, MPCA, and the graph embedding framework the-oretically and derives the incremental learning procedure toadd single sample and multiple samples in detail. The experi-ments on handwritten digit recognition have demonstratedthat ITPCA has achieved better recognition performancethan that of vector-based principal component analysis(PCA), incremental principal component analysis (IPCA),and multilinear principal component analysis (MPCA) algo-rithms. At the same time, ITPCA also has lower time andspace complexity.

In the paper “Optimum performance-based seismic designusing a hybrid optimization algorithm,” S. Talatahari et al.present a hybrid optimization method to optimum seismic

design of steel frames considering four performance levels.These performance levels are considered to determine theoptimum design of structures to reduce the structural cost.A pushover analysis of steel building frameworks subject toequivalent-static earthquake loading is utilized. The algo-rithm is based on the concepts of the charged system search inwhich each agent is affected by local and global best positionsstored in the chargedmemory considering the governing lawsof electrical physics. Comparison of the results of the hybridalgorithm with those of other metaheuristic algorithmsshows the efficiency of the hybrid algorithm.

In the paper entitled “Research on cooperative combat forintegrated reconnaissance-attack-BDA of group LAVs,” L. Binget al. analyze a group system by the theory of Ito stochasticdifferential. The uniqueness and continuity of the solution ofthe system are discussed. Afterwards the model of the systembased on the state transition is established with the finite statemachine automatically. At last, a search algorithm is proposedfor obtaining good feasible solutions for problems. Moreover,simulation results show that model and method are effectivefor dealing with cooperative combat of group LAVs (loiteringair vehicles).

In the paper entitled “Wolf pack algorithm for uncon-strained global optimization,” H.-S. Wu and F.-M. Zhangabstract three intelligent behaviors, scouting, calling, andbesieging, and two intelligent rules, winner-take-all gener-ation rule of lead wolf and stronger-survive renewing ruleof wolf pack. Then they propose a new heuristic swarmintelligent method, named wolf pack algorithm (WPA).Experiments are conducted on a suit of benchmark func-tions with different characteristics, unimodal/multimodal,separable/nonseparable, and the impact of several distancemeasurements and parameters on WPA are discussed. Whatis more, the compared simulation experiments with otherfive typical intelligent algorithms, genetic algorithm, particleswarm optimization algorithm, artificial fish swarm algo-rithm, artificial bee colony algorithm, and firefly algorithm,show that WPA has better convergence and robustness,especially for high-dimensional functions.

In the paper entitled “Towards a unified sentiment lexiconbased on graphics processing units,” L. I. Barbosa-Santillanet al. present an approach to create what they have calleda unified sentiment lexicon (USL). This approach aimsat aligning, unifying, and expanding the set of sentimentlexicons that are available on the web in order to increasetheir robustness of coverage. One problem related to the taskof the automatic unification of different scores of sentimentlexicons is that there are multiple lexical entries for whichthe classification of positive, negative, or neutral (P, N, Z)depends on the unit of measurement used in the annotationmethodology of the source sentiment lexicon. Their USLapproach computes the unified strength of polarity of eachlexical entry based on the Pearson correlation coefficientthat measures how correlated lexical entries are with a valuebetween 1 and −1, where 1 indicates that the lexical entries areperfectly correlated, 0 indicates no correlation, and −1 meansthey are perfectly inversely correlated and so is the unifiedmetrics procedure for CPU and GPU, respectively. Anotherproblem is the high processing time required for computing

Page 10: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

10 Mathematical Problems in Engineering

all the lexical entries in the unification task. Thus, the USLapproach computes a subset of lexical entries in each of the1344GPU cores and uses parallel processing in order to unify155802 lexical entries. The results of the analysis conductedusing the USL approach show that the USL has 95.430 lexicalentries, out of which there are 35.201 considered positive,22.029 negative, and 38.200 neutral. Finally, the runtime is 10minutes for 95.430 lexical entries; this allows a reduction ofthe time computing for the unified metrics by 3 times.

In the paper entitled “Theapplication of FastICA combinedwith related function in blind signal separation,” D. Li etal. propose a new method which is the combination ofrelated function relevance to estimated signal and negativeentropy in fast independent component analysis (FastICA) asobjective function, and the iterative formula is derived with-out any assumptions, then the independent components arefound by maximizing the objective function. The improvedalgorithm shorthand for R-FastICA is applied to extractrandom mixed signals and ventricular late potential (VLP)signal from normal ECG signal; simultaneously the perfor-mance of R-FastICA algorithm is compared with traditionalFastICA through simulation. Experimental results show thatR-FastICA algorithm outperforms traditional FastICA withhigher similarity coefficient and separation precision.

Acknowledgments

We would like to express our gratitude to all of the authorsfor their contributions and the reviewers for their effortproviding constructive comments and feedback. We hopethis special issue offers a comprehensive and timely viewof the area of emerging trends in artificial intelligence andits applications and that it will offer stimulation for furtherresearch.

Yudong ZhangSaeed BalochianPraveen AgarwalVishal Bhatnagar

Orwa Jaber Housheya

Page 11: Editorial Artificial Intelligence and Its Applicationsdownloads.hindawi.com/journals/mpe/2014/840491.pdf · advances in every aspect of articial intelligence technology, including

Submit your manuscripts athttp://www.hindawi.com

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

MathematicsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Mathematical Problems in Engineering

Hindawi Publishing Corporationhttp://www.hindawi.com

Differential EquationsInternational Journal of

Volume 2014

Applied MathematicsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Probability and StatisticsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Journal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Mathematical PhysicsAdvances in

Complex AnalysisJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

OptimizationJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

CombinatoricsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

International Journal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Operations ResearchAdvances in

Journal of

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Function Spaces

Abstract and Applied AnalysisHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

International Journal of Mathematics and Mathematical Sciences

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Algebra

Discrete Dynamics in Nature and Society

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Decision SciencesAdvances in

Discrete MathematicsJournal of

Hindawi Publishing Corporationhttp://www.hindawi.com

Volume 2014 Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Stochastic AnalysisInternational Journal of