Chaotic Time Series Prediction by Fusing Local Methods
|
|
- Maximillian Johnson
- 5 years ago
- Views:
Transcription
1 Chaotic Time Series Prediction by Fusing Local Methods Abstract In this paper, a novel algorithm is proposed to predict chaotic time series. The chaotic time series can be embedded into state space by Takens embedding theorem. The one dimensional data is mapped to a higher dimensional space that provides precise information about the chaotic time series. The upsampling algorithm is used to find more precise nearest neighboring points. Two algorithms which provide accurate results without the knowledge of the underlying dynamics and fuzzy fusion algorithm are employed for one-step and multi-steps ahead forecasting. Simulation results from three typical chaotic time series demonstrate that our method is effective for chaotic time series. Keywords-chaotic time series; fuzzy fusion; embedding theorem; upsampling algorithm; multi-step ahead ; I. INTRODUCTION (HEADING 1) Time series widely exist in our natural world, including macroeconomics, finance, traffic flow, crowd flow, water flow and so on. There are many works focus on modeling time series by linear models, such as auto regressive moving average (ARMA), auto regressive integrated moving average (ARIMA). Strictly speaking, most of the time series can not be accurately depicted by linear models, since the nonlinearity is ubiquitous in the dynamic model. There exists a rich literature for time series. These mothods are classified into two types, linear and nonlinear methods, based on the type of functions that are used in. In linear method, The ARMA method involves two parts, Auto Regression and the Moving Average. It takes into consideration the regression models of data and also the moving average for analyzing the time series data. The ARIMA method is obtained by integrating the ARMA model. Nonlinear method includes radial basis functions, neural networks, and polynomials [1]. The nonlinear methods mainly deal with chaotic time series which is more complex. Chaos theory emerged in 1960s since Lorenz revealed the butterfly effect in studying the weather. After that, chaos have been widely studied and lots of important concepts has been introduced, such as the embedding dimension, embedding delay, chaotic invariants and reconstruction. Application of chaos theory can be found in the area of Yong Wang, Shiqiang Hu* School of Aeronautics and Astronautics Shanghai Jiao Tong University, Shanghai wysjtu2008@gmail.com, sqhu@sjtu.edu.cn production control, human action recognition, signal processing, economic planning and many other fields. The most striking feature of chaotic system is the unpredictability of its feature which is called the sensitivity dependence on initial conditions or butterfly effect. From mining the past to predict the future, the chaotic time series has been widely used in weather forecasting, electric power etc. A number of algorithms have been proposed so far with their own merits and limitations. These methods can be divided into two categories, one-step ahead and multi-step ahead. The later one which is more difficult has to deal with accumulation of errors and uncertainty. For one-step ahead, [6] combined different predictors including the multi-layer perceptron neural network, radial basis function (RBF) neural network, fuzzy inference system, recurrent neural network, Volterra filter, and local linear predictor to predict a chaotic time series. And the linearly constrained least square fusion method is employed to improve the performance of. The multiresolution learning is used in training of predictors. That is signal is decomposed and approximated on different levels of details. And the training process learned from coarse to finest. For multi-step ahead, [8] adopted recursive strategy, which firstly obtained a one-step ahead model and then the estimation value is fed back as an input for the next forecasting. In [9] the authors used a direct strategy that estimated a set of models, each returned a value. There are works that combined the two previous strategies which used different models at each step and the s from previous steps are introduced into the input set. [10] presented an approach for short term for biological and physical phenomenon without knowing the characteristic of the dynamical system. [11] have pointed out that direct strategy is questionable as mapping a p steps into the future would usually be more complicated than recursive strategy. In [2] the authors have shown that recursive strategy performed better on short term. Motivated by the works mentioned above, we use two local algorithms and recursive strategy to forecast chaotic time series. Compared with [6], we make four differences. Firstly, we use two local algorithms. The chaotic * corresponding author address: sqhu@sjtu.edu.cn This paper is jointly supported by the National Natural Science Foundation of China
2 system produces very irregular data which is very different from linear system and weakly nonlinear system. It is not an easy task to build such a model due to the high volatility of the underlying laws behind the chaotic time series. We infer data from nearest neighboring points which is similar to data driven method, hence avoid approximating the function of the dynamical system. Secondly, instead of multiresolution learning, upsampling is employed to interpolate the time series. The higher the upsampling rate, the more detail information of the time series we can obtain, which is especially helpful when the systems fluctuate rapidly. Thirdly, we use fuzzy fusion rule which fuse the two data smoothly. Last but not least, we do not only predict the data one-step ahead, but also solve multi-step ahead. The paper is organized as follows. Section 2 gives a brief background material for state space reconstruction and the method for finding nearest neighboring points. Two algorithms and a fuzzy fusion algorithm are presented in section3. Section 4 gives the experimental results and performance analysis. Finally, conclusions are drawn in section 5. II. BASIC CONCEPTS First let us introduce a few concepts used in our wok. A. state space reconstruction [3] In the time domain, chaotic time series display stochastic behavior. However, it is characterized by the values of a deterministic process in the embedding state space. Given a time series,,,, the state space can be reconstructed according to Takens theorem [7]. Specifically, the time series can be mapped to a higher dimensional state space and hold almost the same structure with the original time series if we select the parameters embedding time delay and embedding dimension d appropriately. Once the two parameters are determined, the state can be written into a matrix. (1) B. Finding nearest neighboring points To find the nearest neighboring points of the state in the reconstructed state space. We first compute the state and its 1 prior state 1,2,, 1 with a predefined metric.,i 1,2,,n1 (2) All the distances are arrayed in an increasing order. And then find the nearest neighboring points. C. Upsampling Finding nearest neighboring points is a critical step for chaotic time series as we need to use the trajectory of neighbors to predict chaotic time series. If the time series is coarsely represented by discrete points, the neighboring point can only be found with coarsely precision. In order to increase the accuracy, interpolation is employed to find the nearest neighboring points with greater precision. The experiments demonstrate that the will be more accurate if the system is interpolated by such way. III. OUR PREDICTION METHODS According to [11] and [12], they have shown that methods based on local mode which presented a number of favorable properties produced more accurate results than other methods. Our local method is based on assumption that equal states have an equal future, and similar state will evolve similarly, at least over short times according to the chaos theory. A. kernel regression algorithm The main idea is to predict the point by using a weighted average of dynamics of neighboring points in the state space. The weights of the neighboring points are defined as: _ (3) _ where _ is the minimum of the neighboring distance, is the number of neighboring points, is predefined parameter, is the ith nearest neighboring point distance. Hence, the next predicted point is given by, (4) x where is one of the K nearest neighboring points of. is the number of neighboring points of. B. Largest Lyapunov Exponent (LLE) algorithm Lyapunov exponent is a dynamical invariant of the attractor, and measures the exponential divergence of the nearby trajectories in the phase space. The LLE depicts the distance of two neighboring points after steps. The distance d between points and its neighboring point is dt xi xt. After steps evolved, the distance between the two points is d t xi pxtp. The relation between and is de, where is LLE. C. Fuzzy fusion rule [13] Fuzzy rule is defined as follows: : if is and is that is, 1,2,, where is the j th fuzzy rule, is the number of fuzzy rule,, and are the input and output of the fuzzy logic respectively., and are fuzzy linguistic term characterized by membership function. In the algorithm, the input of the fuzzy logic is the error between the and the real value. For one-step ahead, errors and in current step between the value and the real value are normalized as e, 708
3 e. The output is the weight in the next step. According to the fuzzy logic rule, data with large error is given smaller weight while data with small error is given larger weight. For multi-step ahead, we use a similar approach. In order to determine the weight of the two algorithms, the real value in the chaotic time series is used to compare with the estimated data. The errors are used in the same way as in one-step ahead to determine the weight of the two results in the next step. D. error analysis In this paper, the relative mean square error (RMSE) is used to evaluate the proposed method. (5) where is th chaotic time series value, is th chaotic time series point real value. is the length of predicted chaotic time series. IV. EXPERIMENTAL RESULTS AND DISCUSSION The experimental results on three different chaotic time series data are presented. A. Ikeda chaotic time series Ikeda chaotic time series is generated by the following equation: x 1cossin (6) y sin t cos (7) where µ 0.9, x 0.1, y 0.1 is the initial value for x and y respectively. The initialization condition is 0.46/1 (8) (9) (10) Integrate the equation (9) and (10) to obtain the following equation, xi11cos 1sin (11) 1 1 (12) In the one-step ahead experiment, we use 400 data for training and 100 for testing, respectively. The errors of the three algorithms are 0.16, and The results are shown in figure 1. The first row is the original data. The second row is the kernel regression algorithm result and errors. The third row is the LLE algorithm result and errors. The fourth row is the fuzzy fusion algorithm result and errors. In the one-step ahead with experiment, we use 200 data for training and 50 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 2. Compared with figure 1 and figure 2, it shows the upsampling algorithm improves the results. Figure 1: Ikeda chaotic time series one-step ahead and errors Figure 2: Ikeda chaotic time series one-step ahead and errors with. In the multi-step ahead experiment, we use 400 data for training and 50 for testing. The errors of the three algorithms are 0.2, 0.3 and 0.22 respectively. The results are shown in figure 3. Figure 3: Ikeda chaotic time series multi-step ahead and errors In the multi-step ahead with upsampling rate 5 experiment, we use 50 data for training and 10 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 4. In the multi-step ahead with upsampling rate 20 experiment, we use 50 data for training and 10 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure
4 Figure 4: Ikeda chaotic time series multi-step ahead and errors with upsampling rate 5. Figure 6: Mackey-Glass chaotic time series and errors In the one-step ahead with upsampling rate 5 experiment, we use 200 data for training and 50 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 7. Figure 5: Ikeda chaotic time series multi-step ahead and errors with upsampling rate 20. Compared with figure 3 and figure 4, we find the improvement of upsampling algorithm is not obvious. However, the performance is greatly promoted in figure5. The RMSE of the three predictors is shown in Table 1. Table 1 Ikeda chaotic time series errors Predictor kernel regression LLE Errors One-step ahead One-step ahead with Multi-step ahead Multi-step ahead with upsampling rate Multi-step ahead with Fusion B. Mackey-Glass chaotic time series The Mackey-Glass chaotic time series is described by the following: x /1 (13) where 0.2, 0.1, 17, In the one-step ahead experiment, we use 400 data for training and 100 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 6. Figure 7: Mackey-Glass chaotic time series one-step ahead and errors with upsampling 5. In the multi-step ahead experiment, we use 600 data for training and 50 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 8. Figure 8: Mackey-Glass chaotic time series multi-step ahead and errors In the multi-step ahead with experiment, we use 400 data for training and 10 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 9. The RMSE of the three predictors is shown in Table
5 Figure 9: Mackey-Glass chaotic time series multi-step ahead and errors with. Table 2 Mackey-Glass chaotic time series errors Predictor kernel regression LLE Fusion Errors One-step ahead One-step ahead with upsampling rate 5 Multi-step ahead Multi-step ahead with C. Lorenz chaotic time series The Lorenz chaotic time series use in this simulation is generated by the following equation: x (14) y (15) z (16) where 16, 45.92, 4, y 1, 0, 1, time step In the one-step ahead experiment, we use 400 data for training and 100 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 10. Figure 11: Lorenz chaotic time series one-step ahead and errors with upsampling 5. In the multi-step ahead experiment, we use 600 data for training and 50 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 12. Figure 12: Lorenz chaotic time series multi-step ahead and errors In the multi-step ahead with experiment, we use 400 data for training and 50 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 13. The RMSE of the three predictors is shown in Table 3. Figure 10: Lorenz chaotic time series and errors In the one-step ahead with upsampling rate 5 experiment, we use 200 data for training and 100 for testing. The errors of the three algorithms are , and respectively. The results are shown in figure 11. Figure 13: Lorenz chaotic time series multi-step ahead and errors with. Table 3 Lorenz chaotic time series errors Predictor kernel regression LLE Fusion Errors One-step ahead
6 One-step ahead with upsampling rate 5 Multi-step ahead Multi-step ahead with D. Discussions: Let us briefly summarize the findings of experiments. (a) In the three experiments, the kernel regression results are generally more accurate than LLE results. One of the reasons is the values of LLE are not precisely determined. There are several methods to estimate the LLE. And each results of the method are different. Another reason is the LLE algorithm is an approximating algorithm. (b) The upsampling step is effective, especially for multi-step ahead. In one-step ahead, the error is small and will be corrected by the real data. In multi-step ahead, no real data for reference. And one feature of the chaotic time series is the butterfly effect mentioned in the first section. Therefore, a small error will be magnified quickly in the multi-step ahead. (c) The experiments show that the bigger the upsampling, the more accurate of the results. Meanwhile, the computation burden will increase. There is no much relation between results and the training data only when the length of training data is a certain large. In the multistep ahead the upsampling algorithm promote the performance significantly. In the signal processing, the Nyquiest theorem [5] points out to reconstruct a signal the sampling frequency should be at least two times higher than the frequency of the original signal. Correspondently, there should be a relationship between the upsampling frequency and the frequency of the chaotic time series. When the upsampling frequency compared with the frequency of the chaotic time series higher than a certain value, the precise nearest neighboring points can be found. The frequency of the Ikeda data is the highest while the Lorenz is the lowest. In the experiments with upsampling, the Ikeda data need higher upsampling frequency than the other two data to maintain the error small. While the of Lorenz data which with lowest frequency among the three data achieve good performance even with low upsampling frequency. (d) In several experiments the results of kernel regression is more accurate than the LLE result and even fusion result. We can observe that the errors of kernel regression generally are much smaller than the errors of LLE. As the fuzzy fusion include the weight of LLE, the performance of fusion algorithm degenerate. It shows that our fusion algorithm may not be the optimal. When we do data fusion, delete the data of large error is better than do directly. This indicates us to find a better criterion to judge the credibility of the fusion data which is out of the scope of this paper. (e) From the one-step ahead, the fusion algorithm can improve the result while in the multi-step ahead. The fusion is also failed as the errors of kernel regression and LLE are large. V. CONCLUSIONS We have presented kernel regression algorithm and LLE algorithm for chaotic time series in this paper. A fusion based on fuzzy rule is also provided. Three chaotic time series are used to verify our algorithm and promising results are achieved. We also compared the upsampling chaotic time series with original data. The results confirm our conjunction that the upsampling algorithm promotes the results. Future works includes finding out proper upsampling rate for multi-step ahead and multivariate chaotic time series which is a more challenging research field. REFERENCES [1] H. Kantz and T. Schreiber, Nonlinear Time Series Analysis, Cambridge University press, [2] Casdagli M. Nonlinear of chaotic time series. Physica D 1989;35: [3] Abarbanel Henry DI. Analysis of observed chaotic data. New York: Springer-Verlag; [4] Abarbanel Henry DI, Brown R, Kadtke JB. Prediction in chaotic nonlinear systems: methods for time series with broadband Fourier spectra. Phys Rev A 1990;41: [5] Monson H. Hayes. Statistical Digital Signal Processing and Modeling. John Wiley & Sons, Inc., [6] M.Aly and H.Leung, Chaotic time series using data fusion, International Conference on Data Fusion (FUSION2001), Montreal, Canada, August [7] Takens F. Detecting strange attractors in turbulence. In: Rand DA, Young LS, editors. Dynamic systems and turbulence, Warwick, Lecture notes in mathematics. Berlin: Springer Verlag; p [8] Haibin Cheng, Pang-Ning Tan, Jing Gao, and Jerry Scripps. Multistepahead time series. In Wee Keong Ng, Masaru Kitsuregawa, Jianzhong Li, and Kuiyu Chang, editors, PAKDD, volume 3918 of Lecture Notes in Computer Science, pages 765{774. Springer, [9] A. Sorjamaa, J. Hao, N. Reyhani, Y. Ji, and A. Lendasse. Methodology for long-term of time series. Neurocomputing, 70(16-18):2861{2869, October [10] G. Sugihara, R.M. May,Nonlinear forecasting as a way of distinguishing chaos from measurement error in time series, Nature 334 (1990) [11] J. Doyne Farmer and John J. Sidorowich. Predicting chaotic time series. Physical Review Letters, 59(8):845{848, August [12] Tim Sauer. Time series by using delay coordinate embedding. In Andreas S. Weigend and Neil A. Gershenfeld, editors, Time Series Prediction, Santa Fe Institue Studies in the Sciences of Complexity, pages Addison-Wesley, [13] T. Takagi and M. Sugeno. Fuzzy identification of systems and its applications to modeling and control. IEEE Transactions on Systems, Man, and Cybernetics, 15(1):116{132,
Time Series Prediction as a Problem of Missing Values: Application to ESTSP2007 and NN3 Competition Benchmarks
Series Prediction as a Problem of Missing Values: Application to ESTSP7 and NN3 Competition Benchmarks Antti Sorjamaa and Amaury Lendasse Abstract In this paper, time series prediction is considered as
More informationarxiv:chao-dyn/ v1 13 Dec 1998
Improving the false nearest neighbors method with graphical analysis T. Aittokallio 1,2, M. Gyllenberg 1, J. Hietarinta 3, T. Kuusela 1,3, T. Multamäki 3. 1 Department of Applied Mathematics, University
More informationFabric Defect Detection Based on Computer Vision
Fabric Defect Detection Based on Computer Vision Jing Sun and Zhiyu Zhou College of Information and Electronics, Zhejiang Sci-Tech University, Hangzhou, China {jings531,zhouzhiyu1993}@163.com Abstract.
More informationTime Series Prediction and Neural Networks
Time Series Prediction and Neural Networks N.Davey, S.P.Hunt, R.J.Frank, University of Hertfordshire Hatfield, UK. Email: {N.Davey, S.P.Hunt, R.J.Frank}@herts.ac.uk Abstract Neural Network approaches to
More informationData Engineering Fuzzy Mathematics in System Theory and Data Analysis
Data Engineering Fuzzy Mathematics in System Theory and Data Analysis Olaf Wolkenhauer Control Systems Centre UMIST o.wolkenhauer@umist.ac.uk www.csc.umist.ac.uk/people/wolkenhauer.htm 2 Introduction General
More informationChaos, fractals and machine learning
ANZIAM J. 45 (E) ppc935 C949, 2004 C935 Chaos, fractals and machine learning Robert A. Pearson (received 8 August 2003; revised 5 January 2004) Abstract The accuracy of learning a function is determined
More informationComparing different interpolation methods on two-dimensional test functions
Comparing different interpolation methods on two-dimensional test functions Thomas Mühlenstädt, Sonja Kuhnt May 28, 2009 Keywords: Interpolation, computer experiment, Kriging, Kernel interpolation, Thin
More informationECM A Novel On-line, Evolving Clustering Method and Its Applications
ECM A Novel On-line, Evolving Clustering Method and Its Applications Qun Song 1 and Nikola Kasabov 2 1, 2 Department of Information Science, University of Otago P.O Box 56, Dunedin, New Zealand (E-mail:
More informationLS-SVM Functional Network for Time Series Prediction
LS-SVM Functional Network for Time Series Prediction Tuomas Kärnä 1, Fabrice Rossi 2 and Amaury Lendasse 1 Helsinki University of Technology - Neural Networks Research Center P.O. Box 5400, FI-02015 -
More informationA Novel Evolving Clustering Algorithm with Polynomial Regression for Chaotic Time-Series Prediction
A Novel Evolving Clustering Algorithm with Polynomial Regression for Chaotic Time-Series Prediction Harya Widiputra 1, Russel Pears 1, Nikola Kasabov 1 1 Knowledge Engineering and Discovery Research Institute,
More informationQUALITATIVE MODELING FOR MAGNETIZATION CURVE
Journal of Marine Science and Technology, Vol. 8, No. 2, pp. 65-70 (2000) 65 QUALITATIVE MODELING FOR MAGNETIZATION CURVE Pei-Hwa Huang and Yu-Shuo Chang Keywords: Magnetization curve, Qualitative modeling,
More informationTWRBF Transductive RBF Neural Network with Weighted Data Normalization
TWRBF Transductive RBF eural etwork with Weighted Data ormalization Qun Song and ikola Kasabov Knowledge Engineering & Discovery Research Institute Auckland University of Technology Private Bag 9006, Auckland
More informationOpen Access Research on the Prediction Model of Material Cost Based on Data Mining
Send Orders for Reprints to reprints@benthamscience.ae 1062 The Open Mechanical Engineering Journal, 2015, 9, 1062-1066 Open Access Research on the Prediction Model of Material Cost Based on Data Mining
More informationSimultaneous Perturbation Stochastic Approximation Algorithm Combined with Neural Network and Fuzzy Simulation
.--- Simultaneous Perturbation Stochastic Approximation Algorithm Combined with Neural Networ and Fuzzy Simulation Abstract - - - - Keywords: Many optimization problems contain fuzzy information. Possibility
More informationSemi-Supervised Clustering with Partial Background Information
Semi-Supervised Clustering with Partial Background Information Jing Gao Pang-Ning Tan Haibin Cheng Abstract Incorporating background knowledge into unsupervised clustering algorithms has been the subject
More informationCLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS
CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS CHAPTER 4 CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS 4.1 Introduction Optical character recognition is one of
More informationNonparametric Regression
Nonparametric Regression John Fox Department of Sociology McMaster University 1280 Main Street West Hamilton, Ontario Canada L8S 4M4 jfox@mcmaster.ca February 2004 Abstract Nonparametric regression analysis
More informationA new fractal algorithm to model discrete sequences
A new fractal algorithm to model discrete sequences Zhai Ming-Yue( 翟明岳 ) a) Heidi Kuzuma b) and James W. Rector b)c) a) School of EE Engineering North China Electric Power University Beijing 102206 China
More informationTime Series Forecasting Methodology for Multiple Step Ahead Prediction
Time Series Forecasting Methodology for Multiple Step Ahead Prediction N. G. Pavlidis, D. K. Tasoulis, M. N. Vrahatis Department of Mathematics, University of Patras Artificial Intelligence Research Center
More informationBasis Functions. Volker Tresp Summer 2017
Basis Functions Volker Tresp Summer 2017 1 Nonlinear Mappings and Nonlinear Classifiers Regression: Linearity is often a good assumption when many inputs influence the output Some natural laws are (approximately)
More informationSparsity Preserving Canonical Correlation Analysis
Sparsity Preserving Canonical Correlation Analysis Chen Zu and Daoqiang Zhang Department of Computer Science and Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China {zuchen,dqzhang}@nuaa.edu.cn
More informationA New Fuzzy Neural System with Applications
A New Fuzzy Neural System with Applications Yuanyuan Chai 1, Jun Chen 1 and Wei Luo 1 1-China Defense Science and Technology Information Center -Network Center Fucheng Road 26#, Haidian district, Beijing
More informationEuropean Journal of Science and Engineering Vol. 1, Issue 1, 2013 ADAPTIVE NEURO-FUZZY INFERENCE SYSTEM IDENTIFICATION OF AN INDUCTION MOTOR
ADAPTIVE NEURO-FUZZY INFERENCE SYSTEM IDENTIFICATION OF AN INDUCTION MOTOR Ahmed A. M. Emam College of Engineering Karrary University SUDAN ahmedimam1965@yahoo.co.in Eisa Bashier M. Tayeb College of Engineering
More informationApplication of Improved Lzc Algorithm in the Discrimination of Photo and Text ChengJing Ye 1, a, Donghai Zeng 2,b
2016 International Conference on Information Engineering and Communications Technology (IECT 2016) ISBN: 978-1-60595-375-5 Application of Improved Lzc Algorithm in the Discrimination of Photo and Text
More informationarxiv: v1 [cond-mat.dis-nn] 30 Dec 2018
A General Deep Learning Framework for Structure and Dynamics Reconstruction from Time Series Data arxiv:1812.11482v1 [cond-mat.dis-nn] 30 Dec 2018 Zhang Zhang, Jing Liu, Shuo Wang, Ruyue Xin, Jiang Zhang
More informationAn Integrated Face Recognition Algorithm Based on Wavelet Subspace
, pp.20-25 http://dx.doi.org/0.4257/astl.204.48.20 An Integrated Face Recognition Algorithm Based on Wavelet Subspace Wenhui Li, Ning Ma, Zhiyan Wang College of computer science and technology, Jilin University,
More informationInput and Structure Selection for k-nn Approximator
Input and Structure Selection for k- Approximator Antti Soramaa ima Reyhani and Amaury Lendasse eural etwork Research Centre Helsinki University of Technology P.O. Box 5400 005 spoo Finland {asorama nreyhani
More informationResearch on Quality Inspection method of Digital Aerial Photography Results
Research on Quality Inspection method of Digital Aerial Photography Results WANG Xiaojun, LI Yanling, LIANG Yong, Zeng Yanwei.School of Information Science & Engineering, Shandong Agricultural University,
More informationImage and Video Quality Assessment Using Neural Network and SVM
TSINGHUA SCIENCE AND TECHNOLOGY ISSN 1007-0214 18/19 pp112-116 Volume 13, Number 1, February 2008 Image and Video Quality Assessment Using Neural Network and SVM DING Wenrui (), TONG Yubing (), ZHANG Qishan
More informationS. Sreenivasan Research Scholar, School of Advanced Sciences, VIT University, Chennai Campus, Vandalur-Kelambakkam Road, Chennai, Tamil Nadu, India
International Journal of Civil Engineering and Technology (IJCIET) Volume 9, Issue 10, October 2018, pp. 1322 1330, Article ID: IJCIET_09_10_132 Available online at http://www.iaeme.com/ijciet/issues.asp?jtype=ijciet&vtype=9&itype=10
More informationHARD, SOFT AND FUZZY C-MEANS CLUSTERING TECHNIQUES FOR TEXT CLASSIFICATION
HARD, SOFT AND FUZZY C-MEANS CLUSTERING TECHNIQUES FOR TEXT CLASSIFICATION 1 M.S.Rekha, 2 S.G.Nawaz 1 PG SCALOR, CSE, SRI KRISHNADEVARAYA ENGINEERING COLLEGE, GOOTY 2 ASSOCIATE PROFESSOR, SRI KRISHNADEVARAYA
More informationA *69>H>N6 #DJGC6A DG C<>C::G>C<,8>:C8:H /DA 'D 2:6G, ()-"&"3 -"(' ( +-" " " % '.+ % ' -0(+$,
The structure is a very important aspect in neural network design, it is not only impossible to determine an optimal structure for a given problem, it is even impossible to prove that a given structure
More informationMODELING FOR RESIDUAL STRESS, SURFACE ROUGHNESS AND TOOL WEAR USING AN ADAPTIVE NEURO FUZZY INFERENCE SYSTEM
CHAPTER-7 MODELING FOR RESIDUAL STRESS, SURFACE ROUGHNESS AND TOOL WEAR USING AN ADAPTIVE NEURO FUZZY INFERENCE SYSTEM 7.1 Introduction To improve the overall efficiency of turning, it is necessary to
More informationCollaborative Rough Clustering
Collaborative Rough Clustering Sushmita Mitra, Haider Banka, and Witold Pedrycz Machine Intelligence Unit, Indian Statistical Institute, Kolkata, India {sushmita, hbanka r}@isical.ac.in Dept. of Electrical
More informationColor Space Projection, Feature Fusion and Concurrent Neural Modules for Biometric Image Recognition
Proceedings of the 5th WSEAS Int. Conf. on COMPUTATIONAL INTELLIGENCE, MAN-MACHINE SYSTEMS AND CYBERNETICS, Venice, Italy, November 20-22, 2006 286 Color Space Projection, Fusion and Concurrent Neural
More informationAutomatic basis selection for RBF networks using Stein s unbiased risk estimator
Automatic basis selection for RBF networks using Stein s unbiased risk estimator Ali Ghodsi School of omputer Science University of Waterloo University Avenue West NL G anada Email: aghodsib@cs.uwaterloo.ca
More informationVector Regression Machine. Rodrigo Fernandez. LIPN, Institut Galilee-Universite Paris 13. Avenue J.B. Clement Villetaneuse France.
Predicting Time Series with a Local Support Vector Regression Machine Rodrigo Fernandez LIPN, Institut Galilee-Universite Paris 13 Avenue J.B. Clement 9343 Villetaneuse France rf@lipn.univ-paris13.fr Abstract
More informationAn intelligent system for automated mathematical modelling and simulation of dynamical engineering systems
An intelligent system for automated mathematical modelling and simulation of dynamical engineering systems O. Castillo a & P. Melin b a Dept. of Computer Science, Instituto Tecnologico de Tijuana, EMail:
More informationFace Hallucination Based on Eigentransformation Learning
Advanced Science and Technology etters, pp.32-37 http://dx.doi.org/10.14257/astl.2016. Face allucination Based on Eigentransformation earning Guohua Zou School of software, East China University of Technology,
More informationLocal Averaging Optimization for Chaotic Time Series Prediction
Local Averaging Optimization for Chaotic Time Series Prediction James McNames Electrical & Computer Engineering, Portland State University, Post Office Box 751, Portland, OR9727-751. Email: mcnames@ee.pdx.edu.
More informationPerformance Degradation Assessment and Fault Diagnosis of Bearing Based on EMD and PCA-SOM
Performance Degradation Assessment and Fault Diagnosis of Bearing Based on EMD and PCA-SOM Lu Chen and Yuan Hang PERFORMANCE DEGRADATION ASSESSMENT AND FAULT DIAGNOSIS OF BEARING BASED ON EMD AND PCA-SOM.
More informationComplexity in Network Economics
Complexity in Network Economics The notion of complexity (in) science has become a bit fuzzy over time. Thus, in this talk I want to shed some light on its meaning, its paradigmatic implication for research
More informationCHAPTER 4 FUZZY LOGIC, K-MEANS, FUZZY C-MEANS AND BAYESIAN METHODS
CHAPTER 4 FUZZY LOGIC, K-MEANS, FUZZY C-MEANS AND BAYESIAN METHODS 4.1. INTRODUCTION This chapter includes implementation and testing of the student s academic performance evaluation to achieve the objective(s)
More information- A Study of Value-Added Tax -The Cases of National Taxation Bureau in the Central Area, Ministry of Finance
2014 6 14-32 - A Study of Value-Added Tax -The Cases of National Taxation Bureau in the Central Area, Ministry of Finance Chong-Si You 100 101 187 145 10 N01 101 99 101 99 1,565,847,055 100 1,703,988,545
More informationRelated Work The Concept of the Signaling. In the mobile communication system, in addition to transmit the necessary user information (usually voice
International Conference on Information Science and Computer Applications (ISCA 2013) The Research and Design of Personalization preferences Based on Signaling analysis ZhiQiang Wei 1,a, YiYan Zhang 1,b,
More informationData mining with Support Vector Machine
Data mining with Support Vector Machine Ms. Arti Patle IES, IPS Academy Indore (M.P.) artipatle@gmail.com Mr. Deepak Singh Chouhan IES, IPS Academy Indore (M.P.) deepak.schouhan@yahoo.com Abstract: Machine
More informationKernel Density Construction Using Orthogonal Forward Regression
Kernel ensity Construction Using Orthogonal Forward Regression S. Chen, X. Hong and C.J. Harris School of Electronics and Computer Science University of Southampton, Southampton SO7 BJ, U.K. epartment
More informationA Random Forest based Learning Framework for Tourism Demand Forecasting with Search Queries
University of Massachusetts Amherst ScholarWorks@UMass Amherst Tourism Travel and Research Association: Advancing Tourism Research Globally 2016 ttra International Conference A Random Forest based Learning
More informationParallel Evaluation of Hopfield Neural Networks
Parallel Evaluation of Hopfield Neural Networks Antoine Eiche, Daniel Chillet, Sebastien Pillement and Olivier Sentieys University of Rennes I / IRISA / INRIA 6 rue de Kerampont, BP 818 2232 LANNION,FRANCE
More informationA Fuzzy C-means Clustering Algorithm Based on Pseudo-nearest-neighbor Intervals for Incomplete Data
Journal of Computational Information Systems 11: 6 (2015) 2139 2146 Available at http://www.jofcis.com A Fuzzy C-means Clustering Algorithm Based on Pseudo-nearest-neighbor Intervals for Incomplete Data
More informationNeural Network based textural labeling of images in multimedia applications
Neural Network based textural labeling of images in multimedia applications S.A. Karkanis +, G.D. Magoulas +, and D.A. Karras ++ + University of Athens, Dept. of Informatics, Typa Build., Panepistimiopolis,
More informationPredictive System for Multivariate Time Series
Predictive System for Multivariate Time Series Jiří Kléma The Gerstner Laboratory for Intelligent Decision Making and Control Faculty of Electrical Engineering Department of Cybernetics Technická 2, Prague
More informationAero-engine PID parameters Optimization based on Adaptive Genetic Algorithm. Yinling Wang, Huacong Li
International Conference on Applied Science and Engineering Innovation (ASEI 215) Aero-engine PID parameters Optimization based on Adaptive Genetic Algorithm Yinling Wang, Huacong Li School of Power and
More informationFuzzy Bidirectional Weighted Sum for Face Recognition
Send Orders for Reprints to reprints@benthamscience.ae The Open Automation and Control Systems Journal, 2014, 6, 447-452 447 Fuzzy Bidirectional Weighted Sum for Face Recognition Open Access Pengli Lu
More informationANALYSIS AND REASONING OF DATA IN THE DATABASE USING FUZZY SYSTEM MODELLING
ANALYSIS AND REASONING OF DATA IN THE DATABASE USING FUZZY SYSTEM MODELLING Dr.E.N.Ganesh Dean, School of Engineering, VISTAS Chennai - 600117 Abstract In this paper a new fuzzy system modeling algorithm
More informationAn Improved KNN Classification Algorithm based on Sampling
International Conference on Advances in Materials, Machinery, Electrical Engineering (AMMEE 017) An Improved KNN Classification Algorithm based on Sampling Zhiwei Cheng1, a, Caisen Chen1, b, Xuehuan Qiu1,
More informationSupport Vector Machines
Support Vector Machines RBF-networks Support Vector Machines Good Decision Boundary Optimization Problem Soft margin Hyperplane Non-linear Decision Boundary Kernel-Trick Approximation Accurancy Overtraining
More informationOptimization Methods for Machine Learning (OMML)
Optimization Methods for Machine Learning (OMML) 2nd lecture Prof. L. Palagi References: 1. Bishop Pattern Recognition and Machine Learning, Springer, 2006 (Chap 1) 2. V. Cherlassky, F. Mulier - Learning
More informationASIAN JOURNAL OF MANAGEMENT RESEARCH Online Open Access publishing platform for Management Research
ASIAN JOURNAL OF MANAGEMENT RESEARCH Online Open Access publishing platform for Management Research Copyright 2010 All rights reserved Integrated Publishing association Review Article ISSN 2229 3795 The
More informationEncoding and Decoding Techniques for Distributed Data Storage Systems
Global Journal of Computer Science and Technology Volume 11 Issue 13 Version 1.0 August 2011 Type: Double Blind Peer Reviewed International Research Journal Publisher: Global Journals Inc. (USA) Online
More informationReal Coded Genetic Algorithm Particle Filter for Improved Performance
Proceedings of 2012 4th International Conference on Machine Learning and Computing IPCSIT vol. 25 (2012) (2012) IACSIT Press, Singapore Real Coded Genetic Algorithm Particle Filter for Improved Performance
More informationA Novel Fault Identifying Method with Supervised Classification and Unsupervised Clustering
A Novel Fault Identifying Method with Supervised Classification and Unsupervised Clustering Tao Xu Department of Automation Shenyang Aerospace University China xutao@sau.edu.cn Journal of Digital Information
More informationA Novel Image Classification Model Based on Contourlet Transform and Dynamic Fuzzy Graph Cuts
Appl. Math. Inf. Sci. 6 No. 1S pp. 93S-97S (2012) Applied Mathematics & Information Sciences An International Journal @ 2012 NSP Natural Sciences Publishing Cor. A Novel Image Classification Model Based
More informationResearch Article Polygon Morphing and Its Application in Orebody Modeling
Mathematical Problems in Engineering Volume 212, Article ID 732365, 9 pages doi:1.1155/212/732365 Research Article Polygon Morphing and Its Application in Orebody Modeling Hacer İlhan and Haşmet Gürçay
More informationAdaptive osculatory rational interpolation for image processing
Journal of Computational and Applied Mathematics 195 (2006) 46 53 www.elsevier.com/locate/cam Adaptive osculatory rational interpolation for image processing Min Hu a, Jieqing Tan b, a College of Computer
More informationTree-based Cluster Weighted Modeling: Towards A Massively Parallel Real- Time Digital Stradivarius
Tree-based Cluster Weighted Modeling: Towards A Massively Parallel Real- Time Digital Stradivarius Edward S. Boyden III e@media.mit.edu Physics and Media Group MIT Media Lab 0 Ames St. Cambridge, MA 039
More informationHead Frontal-View Identification Using Extended LLE
Head Frontal-View Identification Using Extended LLE Chao Wang Center for Spoken Language Understanding, Oregon Health and Science University Abstract Automatic head frontal-view identification is challenging
More informationTime Series Clustering Ensemble Algorithm Based on Locality Preserving Projection
Based on Locality Preserving Projection 2 Information & Technology College, Hebei University of Economics & Business, 05006 Shijiazhuang, China E-mail: 92475577@qq.com Xiaoqing Weng Information & Technology
More informationParameter Estimation in Differential Equations: A Numerical Study of Shooting Methods
Parameter Estimation in Differential Equations: A Numerical Study of Shooting Methods Franz Hamilton Faculty Advisor: Dr Timothy Sauer January 5, 2011 Abstract Differential equation modeling is central
More informationA Texture Feature Extraction Technique Using 2D-DFT and Hamming Distance
A Texture Feature Extraction Technique Using 2D-DFT and Hamming Distance Author Tao, Yu, Muthukkumarasamy, Vallipuram, Verma, Brijesh, Blumenstein, Michael Published 2003 Conference Title Fifth International
More informationIn the Name of God. Lecture 17: ANFIS Adaptive Network-Based Fuzzy Inference System
In the Name of God Lecture 17: ANFIS Adaptive Network-Based Fuzzy Inference System Outline ANFIS Architecture Hybrid Learning Algorithm Learning Methods that Cross-Fertilize ANFIS and RBFN ANFIS as a universal
More informationESTIMATING THE COST OF ENERGY USAGE IN SPORT CENTRES: A COMPARATIVE MODELLING APPROACH
ESTIMATING THE COST OF ENERGY USAGE IN SPORT CENTRES: A COMPARATIVE MODELLING APPROACH A.H. Boussabaine, R.J. Kirkham and R.G. Grew Construction Cost Engineering Research Group, School of Architecture
More informationChaotic Time Series Prediction Using Combination of Hidden Markov Model and Neural Nets
Chaotic Time Series Prediction Using Combination of Hidden Markov Model and Neural Nets Saurabh Bhardwaj, Smriti Srivastava, Member, IEEE, Vaishnavi S., and J.R.P Gupta, Senior Member, IEEE Netaji Subhas
More informationFuzzy if-then rules fuzzy database modeling
Fuzzy if-then rules Associates a condition described using linguistic variables and fuzzy sets to a conclusion A scheme for capturing knowledge that involves imprecision 23.11.2010 1 fuzzy database modeling
More informationColor-Based Classification of Natural Rock Images Using Classifier Combinations
Color-Based Classification of Natural Rock Images Using Classifier Combinations Leena Lepistö, Iivari Kunttu, and Ari Visa Tampere University of Technology, Institute of Signal Processing, P.O. Box 553,
More informationAPPLICATION OF RADON TRANSFORM IN CT IMAGE MATCHING Yufang Cai, Kuan Shen, Jue Wang ICT Research Center of Chongqing University, Chongqing, P.R.
APPLICATION OF RADON TRANSFORM IN CT IMAGE MATCHING Yufang Cai, Kuan Shen, Jue Wang ICT Research Center of Chongqing University, Chongqing, P.R.China Abstract: When Industrial Computerized Tomography (CT)
More informationAppART + Growing Neural Gas = high performance hybrid neural network for function approximation
1 AppART + Growing Neural Gas = high performance hybrid neural network for function approximation Luis Martí Ý Þ, Alberto Policriti Ý, Luciano García Þ and Raynel Lazo Þ Ý DIMI, Università degli Studi
More informationIntelligent Methods in Modelling and Simulation of Complex Systems
SNE O V E R V I E W N OTE Intelligent Methods in Modelling and Simulation of Complex Systems Esko K. Juuso * Control Engineering Laboratory Department of Process and Environmental Engineering, P.O.Box
More informationEnhancing Forecasting Performance of Naïve-Bayes Classifiers with Discretization Techniques
24 Enhancing Forecasting Performance of Naïve-Bayes Classifiers with Discretization Techniques Enhancing Forecasting Performance of Naïve-Bayes Classifiers with Discretization Techniques Ruxandra PETRE
More informationApplication of Geometry Rectification to Deformed Characters Recognition Liqun Wang1, a * and Honghui Fan2
6th International Conference on Electronic, Mechanical, Information and Management (EMIM 2016) Application of Geometry Rectification to Deformed Characters Liqun Wang1, a * and Honghui Fan2 1 School of
More informationUsing Local Trajectory Optimizers To Speed Up Global. Christopher G. Atkeson. Department of Brain and Cognitive Sciences and
Using Local Trajectory Optimizers To Speed Up Global Optimization In Dynamic Programming Christopher G. Atkeson Department of Brain and Cognitive Sciences and the Articial Intelligence Laboratory Massachusetts
More informationA Combined Encryption Compression Scheme Using Chaotic Maps
BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 13, No 2 Sofia 2013 Print ISSN: 1311-9702; Online ISSN: 1314-4081 DOI: 10.2478/cait-2013-0016 A Combined Encryption Compression
More informationPolynomial Curve Fitting of Execution Time of Binary Search in Worst Case in Personal Computer
Polynomial Curve Fitting of Execution Time of Binary Search in Worst Case in Personal Computer Dipankar Das Assistant Professor, The Heritage Academy, Kolkata, India Abstract Curve fitting is a well known
More informationTable of Contents. Recognition of Facial Gestures... 1 Attila Fazekas
Table of Contents Recognition of Facial Gestures...................................... 1 Attila Fazekas II Recognition of Facial Gestures Attila Fazekas University of Debrecen, Institute of Informatics
More information^ Springer. Computational Intelligence. A Methodological Introduction. Rudolf Kruse Christian Borgelt. Matthias Steinbrecher Pascal Held
Rudolf Kruse Christian Borgelt Frank Klawonn Christian Moewes Matthias Steinbrecher Pascal Held Computational Intelligence A Methodological Introduction ^ Springer Contents 1 Introduction 1 1.1 Intelligent
More informationVideo Inter-frame Forgery Identification Based on Optical Flow Consistency
Sensors & Transducers 24 by IFSA Publishing, S. L. http://www.sensorsportal.com Video Inter-frame Forgery Identification Based on Optical Flow Consistency Qi Wang, Zhaohong Li, Zhenzhen Zhang, Qinglong
More informationAdaptive Zoom Distance Measuring System of Camera Based on the Ranging of Binocular Vision
Adaptive Zoom Distance Measuring System of Camera Based on the Ranging of Binocular Vision Zhiyan Zhang 1, Wei Qian 1, Lei Pan 1 & Yanjun Li 1 1 University of Shanghai for Science and Technology, China
More informationModeling with Uncertainty Interval Computations Using Fuzzy Sets
Modeling with Uncertainty Interval Computations Using Fuzzy Sets J. Honda, R. Tankelevich Department of Mathematical and Computer Sciences, Colorado School of Mines, Golden, CO, U.S.A. Abstract A new method
More informationSOM+EOF for Finding Missing Values
SOM+EOF for Finding Missing Values Antti Sorjamaa 1, Paul Merlin 2, Bertrand Maillet 2 and Amaury Lendasse 1 1- Helsinki University of Technology - CIS P.O. Box 5400, 02015 HUT - Finland 2- Variances and
More informationAn Empirical Study of Lazy Multilabel Classification Algorithms
An Empirical Study of Lazy Multilabel Classification Algorithms E. Spyromitros and G. Tsoumakas and I. Vlahavas Department of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
More informationFACE RECOGNITION USING FUZZY NEURAL NETWORK
FACE RECOGNITION USING FUZZY NEURAL NETWORK TADI.CHANDRASEKHAR Research Scholar, Dept. of ECE, GITAM University, Vishakapatnam, AndraPradesh Assoc. Prof., Dept. of. ECE, GIET Engineering College, Vishakapatnam,
More informationMultidirectional 2DPCA Based Face Recognition System
Multidirectional 2DPCA Based Face Recognition System Shilpi Soni 1, Raj Kumar Sahu 2 1 M.E. Scholar, Department of E&Tc Engg, CSIT, Durg 2 Associate Professor, Department of E&Tc Engg, CSIT, Durg Email:
More information99 International Journal of Engineering, Science and Mathematics
Journal Homepage: Applications of cubic splines in the numerical solution of polynomials Najmuddin Ahmad 1 and Khan Farah Deeba 2 Department of Mathematics Integral University Lucknow Abstract: In this
More informationBasis Functions. Volker Tresp Summer 2016
Basis Functions Volker Tresp Summer 2016 1 I am an AI optimist. We ve got a lot of work in machine learning, which is sort of the polite term for AI nowadays because it got so broad that it s not that
More informationGeneric Face Alignment Using an Improved Active Shape Model
Generic Face Alignment Using an Improved Active Shape Model Liting Wang, Xiaoqing Ding, Chi Fang Electronic Engineering Department, Tsinghua University, Beijing, China {wanglt, dxq, fangchi} @ocrserv.ee.tsinghua.edu.cn
More informationFace Recognition by Combining Kernel Associative Memory and Gabor Transforms
Face Recognition by Combining Kernel Associative Memory and Gabor Transforms Author Zhang, Bai-ling, Leung, Clement, Gao, Yongsheng Published 2006 Conference Title ICPR2006: 18th International Conference
More informationAn Autoassociator for Automatic Texture Feature Extraction
An Autoassociator for Automatic Texture Feature Extraction Author Kulkarni, Siddhivinayak, Verma, Brijesh Published 200 Conference Title Conference Proceedings-ICCIMA'0 DOI https://doi.org/0.09/iccima.200.9088
More informationA Data Classification Algorithm of Internet of Things Based on Neural Network
A Data Classification Algorithm of Internet of Things Based on Neural Network https://doi.org/10.3991/ijoe.v13i09.7587 Zhenjun Li Hunan Radio and TV University, Hunan, China 278060389@qq.com Abstract To
More informationA derivative-free trust-region algorithm for reliability-based optimization
Struct Multidisc Optim DOI 10.1007/s00158-016-1587-y BRIEF NOTE A derivative-free trust-region algorithm for reliability-based optimization Tian Gao 1 Jinglai Li 2 Received: 3 June 2016 / Revised: 4 September
More informationChapter 7. Conclusions and Future Work
Chapter 7 Conclusions and Future Work In this dissertation, we have presented a new way of analyzing a basic building block in computer graphics rendering algorithms the computational interaction between
More information