editorial recent advancements in signal processing and...

EditorialRecent Advancements in Signal Processing andMachine Learning

Gelan Yang,1 Su-Qun Cao,2 and Yue Wu3

1 College of Information Science and Engineering, Hunan City University, Yiyang 413000, China2 Faculty of Mechanical Engineering, Huaiyin Institute of Technology, Huai’an 223003, China3Department of Electrical and Computer Engineering, Tufts University, Medford, MA 02155, USA

Correspondence should be addressed to Gelan Yang; [email protected]

Received 26 March 2014; Accepted 26 March 2014; Published 14 April 2014

Copyright © 2014 Gelan Yang 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.

Recent advances of consumer electronics, like iphone 5s,google glasses, Xbox Kinect, and so forth, bring peoplerevolutionary experiences of human-machine interactions.Behind these innovations are the successes of variousrecent cutting-edge technologies, including voice recogni-tion, object recognition, and motion recognition. Thoughmany of these technologies are still far from perfect, theseexamples demonstrate the usefulness and importance ofsignal processing and machine learning (SPML) research.

SPML, however, clearly goes far beyond these well-known recognition technologies and roots deeply in manyaspects of academic research and industrial development.Indeed, wherever digital sensors require signal processing,and wherever decision-making problems can be formed ina machine learning manner. Moreover, SPML is also deeplyinvolved: on one hand, ML could take digital signals as rawfeatures to learn rules; on the other hand, many statisticalSP techniques are essentially ML solutions. For example,wavelet transform proposed for SP now has been widely usedas a preprocessing for many machine learning applications,while probabilistic graphic models often used in an expertsystem now show their potentials in image segmentation andrecognition.

For these reasons, the aim of this special issue is toconsider the recent advancements in SPML together. In total,this special issue contains thirty-four papers studying variousSPML problems including image registration, pattern recog-nition, texture analysis, surveillance, biometrics, human-machine interface, image/video compression/encryption,image enhancement, and face recognition. A summarizeddescription of these papers is given below.

“Strategies for exploiting independent cloud implementa-tions of biometric experts in multibiometric scenarios” by P.Peer et al. presents an analysis of different strategies forcombining independent cloud implementations of biometricexperts into a multibiometric recognition system. Analysisresults suggest that fixed fusion rules combining single expertsystems at the matching score level are the most suitablefor the studied task as they provide a good balance amongexpected performance gains and other important factors.

“Linear chromatic adaptation transform based on Delau-nay triangulation” by R. Kreslin et al. suggests amethod usingDelaunay triangulation and linear transformations in orderto transform the input image. It relies on the color values ofthe patches of the Macbeth color checker captured under thesame illuminant as the input imagemust be known.Objectiveevaluation showed that the proposed method outperformsexisting CAT methods by more than 21%, which performsstatistically significantly better than that of other existingmethods.

“Language recognition using latent dynamic conditionalrandom field model with phonological features” by S. Boonsuket al. proposes an acoustic SLR system based on the latent-dynamic conditional random field (LDCRF) model usingphonological features (PFs). Evaluated on the NIST LRE2007 corpus, the proposed method showed an improvementover the baseline systems. Additionally, it showed comparableresult with the acoustic system based on i-vector. Thisresearch demonstrates that utilizing the PF attributes as amean to integrate linguistic information with the acousticapproach can enhance the performance.

Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2014, Article ID 549024, 4 pageshttp://dx.doi.org/10.1155/2014/549024

2 Mathematical Problems in Engineering

“Human skeleton model based dynamic features for walk-ing speed invariant gait recognition” by J. Kovac et al. proposesa skeleton model based gait recognition system focusing onmodelling gait dynamics and eliminating the influence ofsubject appearance on recognition. Furthermore, this papertackles the problem of walking speed variation and proposesspace transformation and feature fusion that mitigates itsinfluence on recognition performance.

“Efficient LED-SAC sparse estimator using fast sequentialadaptive coordinate-wise optimization (LED-2SAC)” by T.Yousefi Rezaii et al. presents a novelmethod benefits from thesparsity of the signal both in the optimization criterion (LED)and its solution path, denoted by Sparse SAC (2SAC). Thenew reconstruction method is consequently more efficientand considerably faster compared to the LED-SACalgorithm,in terms of adaptability and convergence rate with a muchlower computational complexity.

“A joint learning approach to face detection inwavelet com-pressed domain” by S.-H. Huang et al. carries out a novel facedetection system working directly in the wavelet compresseddomain. This methodology involves a feature space warpingprocess, a paired feature learning scheme, an ID3-like jointfeature plane quantization method, and a weak Bayesianclassifier. Experimental results on the benchmarking facedatasets showed that the proposed face detection systemworking in the compressed domain achieves similar accuracyto that of Viola and Jones’ face detector.

“Improvement and simulation of an autonomous time syn-chronization algorithm for a layered satellite constellation” byF.Huang et al. investigates the autonomous time synchroniza-tion algorithm that corresponds to the layered constellationstructure, analyzes themain error of the time synchronizationalgorithm, and proposesmethods to improve the characteris-tics of satellite movement in the constellation. Its simulationresults show that in a condition with simulation errors, thetime synchronization precision of this improved algorithmcan be controlled within 5 ns and used in high-precisionautonomous time synchronization between layered satelliteconstellations.

“The new mathematical model of motion compensationfor stepped-frequency radar signal” by Y. Lin et al. presentsa novel mathematical method to estimate target speed forthe stepped-frequency radar. Its numeric simulation resultsconfirm that this new method is effective and predominantin terms of much higher estimation accuracy in a low SNRand much larger estimation range of target speed.

“Human walking pattern recognition based on KPCA andSVM with ground reflex pressure signal” by Z. Peng et al.investigates an algorithm based on the ground reflex pressure(GRF) signal obtained from a pair of sensing shoes for humanwalking pattern recognition. Experimental results showedthat algorithm fusing SVM and KPCA had better recognitionperformance.

“Application of global optimization methods for featureselection and machine learning” by S. Wu et al. proposes anovel immune clonal genetic algorithm for the feature selec-tion problem. The proposed algorithm largely simplifies thefeature selection process without trading off its effectiveness.

It shows higher classification accuracy than compared featureselection algorithms.

“An initial value calibration method for the wheel forcetransducer based on memetic optimization framework” by G.Lin et al. proposes an automatic solution without additionalcalibration equipment or manual operation. In this method,a vehicle with the wheel force transducer (WFT) is drivenon a flat road with a constant speed. A real WFT data isused to verify the proposed method and result shows thatit is superior to traditional solutions and can improve themeasurement accuracy effectively.

“A routing algorithm for WiFi-based wireless sensor net-work and the application in automatic meter reading” by L.Li et al. introduces a new architecture of WiFi-based wirelesssensor network, which is suitable for the next generationAMR system. It also proposes a new improved routingalgorithm called energy saving-based hybrid wireless meshprotocol (E-HWMP).

“An efficient web usage mining approach using chaos opti-mization and particle swarm optimization algorithm based onoptimal feedback model” by L. Dai et al. proposes an efficientparticle swarm chaos optimization mining algorithm. It usesa user feedback model to provide a listing of best-matchingwebpages for user. Its test results show that this approachsignificantly outperforms other algorithms in the aspects ofresponse time, execution time, precision, and recall.

“Video shot boundary recognition based on adaptive local-ity preserving projections” by Y. Xiao et al. introduces a novelvideo shot boundary recognitionmethod. It firstly defines thediscriminating similarity with mode prior probabilities andan adaptive neighborhood selection strategy and then uses anoptimized multiple kernel support vector machine to classifyvideo frames into boundary and nonboundary frames.

“A virtual channels scheduling algorithmwith broad appli-cability based on movable boundary” by Y. Tian et al. presentsa novel algorithm for virtual channel scheduling based onmovable boundary. It divides slots into synchronous onesand asynchronous ones, reduces the scheduling delay, andimproves the channel utilization ratio. The proposed methodoutperforms the DSA algorithm, when the time delay andapplicability scope are concerned. In addition, the algorithmis suitable for the scheduling of diverse data sources.

“Comprehensive models for evaluating rock mass stabilitybased on statistical comparisons of multiple classifiers” by L.Dong and X. Li demonstrates the applicability and feasibilityof RF, SVM, Bayes (NBC), Fisher, LR, and NN classificationmodels to evaluate the rockmass stability of slope. Resultsshow that the established RF, SVM, Bayes, Fisher, LR, andNNclassification models can evaluate the slope status with a highaccuracy.

“Unsupervised optimal discriminant vector based featureselection method” by S.-Q. Cao and J. H. Manton proposesan efficient unsupervised feature selection method based onunsupervised optimal discriminant vector. It aims to findimportant features without using class labels. Two exper-iments on Wine dataset and fault diagnosis demonstratethat the proposed method is able to find important featuresand is a reliable and efficient feature selection methodologycompared to SUD and Relief-F methods.

Mathematical Problems in Engineering 3

“Penalized maximum likelihood algorithm for positronemission tomography by using anisotropic median-diffusion”byQ.He and L.Huang presents an approach to improving thequality of positron emission tomography images. By fusing ananisotropic median-diffusion filter to maximum-likelihoodexpectation-maximization algorithm, testing results demon-strate high-quality image reconstruction and denoising withbetter edge preserving capacities.

“A novel fusion method by static and moving facialcapture” by S. Liu et al. introduces a fusion facial detectionmethod in moving environment in the field of face recogni-tion. Experimental results show that this method has betterrobustness and accuracy.

“Lower power design for UHF RF CMOS circuits based onthe power consumption acuity” by N. Xiang-jie and L. Huaconveys a lower power design for UHF RF CMOS circuitbased on the power consumption acuity due to the excessiveenergy consumption of UHF tag.The simulation results showthat the leakage power of rectifier in this method is obviouslyless than the conventional rectifier. The proposed designmethod is suitable for various industrial productions, publicmanagement, and daily life use.

“Accurate counting bloom filters for large-scale data pro-cessing” by W. Li et al. proposes a multilevel optimizationapproach to build an accurate counting bloom filter (ACBF)for reducing the false positive probability. Experiments onrealistic datasets show that ACBF can greatly reduce the falsepositive probability as well as the map outputs. Meanwhile,compared to the classic solution, this method also improvesthe join execution times by 20%.

“A heuristic feature selection approach for text categoriza-tion by using chaos optimization and genetic algorithm” by H.Chen et al. proposes a novel text categorization algorithmcalled chaos genetic feature selection optimization. Experi-mental results show that the proposed algorithm effectivelysimplifies the feature selection process which can also obtainhigher classification accuracy with a smaller feature set.

“A simple and high performing rate control initializationmethod for H.264 AVC coding based on motion vector mapand spatial complexity at low bitrate” by Y. Wu and S.-W. Kodescribes a simple and high performance initial QP deter-mining method based on motion vector map. Simulationresults indicate that this algorithm outperforms conventionalmethods under many objective and subjective criteria.

“Subband adaptive filtering with 𝑙1-norm constraint for

sparse system identification” by Y.-S. Choi introduces a nor-malized subband adaptive filter NSAF integrating a newweighted 𝑙

1-normconstraint.Numerical results prove that the

𝑙1-norm regularized NSAFs outperform the classical NSAF

solutions, especially for identifying a sparse system.“Image encryption using the chaotic Josephusmatrix” byG.

Yang et al. introduces a new image encryption solution usingthe chaotic Josephusmatrix. It extends the conventional Jose-phus traversing to a matrix form and proposes a treatmentto improve the randomness of this matrix by mixing chaoticmaps. The proposed CJPM is parametric and is uniquelydependent on the set of parameters, which is sufficientlylarge to provide a secure size of key space. Simulation resultsdemonstrate that an encrypted image of using this method

is very random-like from the perspective of human visualinspection.

“Active semisupervised clustering algorithm with labelpropagation for imbalanced and multidensity datasets” byM. Leng et al. provides an active semisupervised clusteringalgorithm based on active data selection and semisuper-vised clustering algorithm on multidensity and imbalanceddatasets. Testing results show that the proposed semisuper-vised clustering has higher accuracy and stable performancecompared to other clustering and semisupervised clusteringalgorithms, especially in case of the datasets are multidensityand imbalanced.

“A novel machine learning strategy based on two-dimensional numerical models in financial engineering” byQ. Xu puts forward a two-dimensional numerical model formachine learning to simulate major US stock market index,which uses a nonlinear implicit finite-difference method tofind numerical solutions of the two-dimensional simulationmodel. For the purpose of better prediction of the futuretrend of the index, experimental results show that theproposed algorithm reduces the prediction error and improveforecasting precision.

“Analysis and denoising of hyperspectral remote sensingimage in the curvelet domain” by D. Xu et al. provides anew denoising algorithm based on the characteristics ofhyperspectral remote sensing image (HRSI) in the curveletdomain. The detailed subband images in the same scaleand same direction from different wavelengths of HRSI arestacked to obtain new 3D datacubes of the curvelet domain.Themultiple linear regressionmethod is also introduced.Thesimulated data experimental results show that the proposedalgorithm is superior to the compared algorithms in thereference in terms of SNR.

“Efficient interaction recognition through positive actionrepresentation” by T. Hu et al. proposes a novel approach todecompose two-person interaction into a positive action anda negative action for more efficient behavior recognition. Inthis way, interaction recognition can be simplified to positiveaction-based recognition. Also, they created a new datasetwith six types of complex human interaction. Experimentalresults showed that the proposed recognition technique ismore accurate than the traditional method, shortens thesample training time, and hence achieves comprehensivesuperiority.

“Tensorial kernel principal component analysis for actionrecognition” by C. Liu et al. proposes a novel tensorial kernelprincipal component analysis (TKPCA) for feature extractionfromaction objects, which extends the conventional principalcomponent analysis (PCA) in two perspectives: workingdirectly with multidimensional data (tensors) in their nativestate and generalizing an existing linear technique to itsnonlinear version by applying the kernel trick. Experimentswith real action datasets show that the proposed method isinsensitive to both noise and occlusion and performs wellcompared with state-of-the-art algorithms.

“Low-Complexity compression algorithm for hyperspectralimages based on distributed source coding” by Y. Nian etal. proposes distributed compression algorithm for realizing

4 Mathematical Problems in Engineering

both lossless and lossy compression. It is implemented by per-forming scalar quantization strategy on the original hyper-spectral images followed by distributed lossless compression.Multilinear regressionmodel is introduced for improving thequality of side information. Experimental results show thatthe compression performance of the proposed algorithm iscompetitive with that of state-of-the-art compression algo-rithms for hyperspectral images.

“Model for recognizing key factors and applications thereofto engineering” by B. Shi and G. Chi presents a model forrecognizing key factors while using collinearity diagnosticsfor deleting factors of repeated information and logisticregression for selecting factors. Experimental results from2044 observations in finical engineering show that the 13indicators are recognized as key factors to distinguish thegood customers from the bad customers and thus demon-strate the effectiveness of the proposed method.

“Application of fuzzy set theory to quantitative analysisof correctness of the mathematical model based on the ADImethod during solidification” by X. Niu et al. proposes amodel based on the equivalent specific heat method andthe ADI method aiming at improving the computationalefficiency. Experimental results show that for a thick-walledposition, the time step influences the simulation results of thetemperature field and the number of casting meshes has littleinfluence on the simulation results of temperature field, whilefor a thin-walled position a larger influence exists.

“Minimum error thresholding segmentation algorithmbased on 3D grayscale histogram” by J. Liu et al. proposesa novel algorithm called three-dimensional minimum errorthresholding (3D-MET) according to the relative entropytheory. The 3D histogram is obtained by combining rawinformation of pixel intensity distribution and relevant infor-mation of neighboring pixels within an image. Experimentalresults indicate that the proposed approach provides superiorsegmentation performance compared to other methods forgray image segmentation.

Acknowledgments

Thanks are due to the authors of the special issue for theircontributions and thanks to the reviewers for their valuablecomments on the submissions. Gelan Yang acknowledges thesupport by Scientific Research Fund of Hunan ProvincialEducation Department (Grant no. 12B023). Su-Qun Caoacknowledges the support by the National Spark Plan ofChina (Grant no. 2013GA690404).

Gelan YangSu-Qun Cao

Yue Wu

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

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

MathematicsJournal of


Mathematical Problems in Engineering

Hindawi Publishing Corporationhttp://www.hindawi.com

Differential EquationsInternational Journal of

Volume 2014

Applied MathematicsJournal of


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

Journal of


Mathematical PhysicsAdvances in

Complex AnalysisJournal of


OptimizationJournal of


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

International Journal of


Operations ResearchAdvances in

Journal of


Function Spaces

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

International Journal of Mathematics and Mathematical Sciences


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


Algebra

Discrete Dynamics in Nature and Society



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

editorial recent advancements in signal processing and...

Documents