By Topic

Machine Learning and Cybernetics, 2002. Proceedings. 2002 International Conference on

Date 4-5 Nov. 2002

Go

Filter Results

Displaying Results 1 - 25 of 119
  • 2002 International Conference on Machine Learning and Cybernetics [front matter]

    Page(s): 0_1 - 0_23
    Save to Project icon | Request Permissions | PDF file iconPDF (1004 KB)  
    Freely Available from IEEE
  • Author index

    Page(s): 2257 - 2265
    Save to Project icon | Request Permissions | PDF file iconPDF (528 KB)  
    Freely Available from IEEE
  • Locating essential facial features using neural visual model

    Page(s): 1914 - 1919 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (424 KB) |  | HTML iconHTML  

    Facial feature detection plays an important role in applications such as human computer interaction, video surveillance, face detection and face recognition. We propose a facial feature detection algorithm for all types of face images in the presence of several image conditions. There are two main step: the facial feature extraction from original face image, and the coverage of the features by rectangular blocks. A neural visual model (NVM) is used to recognize all possibilities of facial feature positions for the first step. Input parameters are obtained from the face characteristics and the positions of facial features not including any intensity information. For the better results, some incorrect decisions of facial feature positions are improved by image processing technique called dilation. Our algorithm is successfully tested with various types of faces which are color images, gray images, binary images, wearing the sunglasses, wearing the scarf, lighting effect, noise and blurring images, color and sketch images from animated cartoon. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Construction of orthogonal multiwavelet

    Page(s): 1968 - 1972 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (316 KB) |  | HTML iconHTML  

    For compactly supported m-band (m≥2, m∈Z) orthonormal multiwavelet systems, an algorithm to construct compactly supported orthogonal multiwavelets from the associated multi-scaling functions is presented in this paper. This method is simple for computation, does not need to solve a set of nonlinear equations in unknown matrices or factorize a polynomial matrix into a special form, and is not restrained by the multiplicity of multiwavelets. It only needs to solve some linear equations and compute some matrices. As an example, the GHM multiwavelets are derived via this method. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Oscillation suppression for a class of flexible system with input shaping technique

    Page(s): 2108 - 2111 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (324 KB) |  | HTML iconHTML  

    The input shaping method has been applied to control many types of flexible systems. Input shaping is a simple and effective method. for reducing the residual vibration when positioning lightly damped systems. Also, it can be used to a class of system with oscillation to reduce or even eliminate the overshoot with the rise time as short as possible. In this paper, several types of command shaping techniques are introduced and compared. The relationship between different kinds of input shapers can be seen as well as their characteristics. Finally, applying the traditional ZVD method and the OATF to shape a closed-loop system verifies the proposed ideas. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A retrieval model in multiple level image information

    Page(s): 2055 - 2058 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (320 KB) |  | HTML iconHTML  

    In this paper, the features of image are analyzed, and a multiple granularity hierarchy description model is proposed. Based on the model, a hierarchy image retrieval model is proposed. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Discovery of emerging patterns from nearest neighbors

    Page(s): 1920 - 1925 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (440 KB) |  | HTML iconHTML  

    In this paper, we propose a scalable classifier that uses jumping emerging patterns (JEPs), which are combinations of values that occur in one class. The original classifier, DeEPs, is an instance-based classifier that operates on all instances in real-time. It discovers maximal patterns that occur throughout the entire database and identifies JEPs by using these patterns. The necessary computational effort, though, is likely to increase when DeEPs is applied to a large database. Our proposed classifier operates on the nearest neighbors of a test instance. This reduction of instances improves scalability as the database volume increases. Moreover, our classifier imposes a restriction regarding JEPs discovery, so that it excludes patterns that cannot be identified as either correct JEPs or JEPs caused by the maximal patterns missing from nearest neighbors. These probably incorrect JEPs are specialized with additional items and participate in class determination. Our classifier perform significantly faster with these two enhancements, while it remains as accurate as the original classifier. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Mining the weights of similarity measure through learning

    Page(s): 1837 - 1841 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (558 KB) |  | HTML iconHTML  

    An approach is proposed to minimize a fuzzy feature evaluation index function by genetic algorithms. Since not all evaluation indexes perform well, a cross-entropy is introduced to measure the fuzziness of the evaluation function. Experimental results show that with the function of cross-entropy, a suitable evaluation index is chosen, the fuzziness is reduced and the corresponding clustering is optimized. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A decoupling control method with improving genetic algorithm

    Page(s): 2112 - 2115 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (306 KB) |  | HTML iconHTML  

    A decoupling control method of genetic algorithm is presented on a basis. of improving fitness function. Moreover, this method is,validated by using analysis, simulations and experiments. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Experimental verification for zero power control of 0.5KWh class flywheel system using magnetic bearing with gyroscopic effect

    Page(s): 2059 - 2062 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (305 KB) |  | HTML iconHTML  

    This paper proposes a new method on how to construct the reduced-order model of rotor system with, gyroscopic effect used for control system design of closed loop system. In order to demonstrate the validity of this method, the controlled model and zero power control of a 0.5KWh class flywheel system using magnetic bearing with gyroscopic effect are given in this paper. The proposed method is verified by experiment in this paper. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • DFL nonlinear control design and its application in water-level nonlinear plant

    Page(s): 1774 - 1777 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (267 KB) |  | HTML iconHTML  

    Designs of adaptive controllers to improve the performance of water-level nonlinear plants have been a topic of research for a long time. Few papers, however, have considered both the stability enhancement and speed increment of the dynamic response of the plant using adaptive controllers. The aim of this paper is to present the design of "direct feedback linearization" (DFL), which can solve the problems related to the nonlinear characteristic of the plant, and to apply the DFL method to a water-level nonlinear plant to improve its performance. The real time control experiments show that the DFL which linearizes the water-level nonlinear plant is valid in the large range, while the classical linearization method is valid only in the small range. Under the precondition of no steady-state error, there is no peak overshoot. Compared with the classical PID control method, the DFL method gives a satisfactory result. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The subsystem of the fuzzed rough sets based on equivalence class

    Page(s): 2157 - 2159 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (259 KB) |  | HTML iconHTML  

    A method for rough sets fuzzification by using the equivalence relation is given. The definitions of the set of equivalence class and rough set being fuzzified are presented and the fact that fuzzified rough sets can constitute the fuzzy subsystem is proved. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An optimal point in scheduling real-time tasks process based on fault tolerant imprecise computation model

    Page(s): 2063 - 2068 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (390 KB) |  | HTML iconHTML  

    Fault tolerance is an important issue due to the critical nature of the supported tasks of real-time computer systems, since timing constraints must not be violated. The imprecise computation technique has been proposed as a way to handle transient overload and to enhance fault tolerant of real-time systems. This paper introduces an exact theoretical analysis for the imprecise computation model based on three principles of maximize reward-based test, minimize response-time test, and minimize errors test, then finds optimal-point in scheduling process to satisfy three scheduling conditions. Further this is also demonstrated by the simulation results. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A new algorithm of selecting the radial basis function networks center

    Page(s): 1801 - 1804 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (310 KB) |  | HTML iconHTML  

    The selecting of the radial basis function center is a key factor that influences the performance of networks. In this paper we first introduce briefly the fuzzy c-mean algorithm and k-nearest-neighbor algorithm as to the selection of the radial basis function center, and then we present a δ-nearest-neighbor cluster algorithm which combines the k-nearest-neighbor algorithm with fuzzy c-mean algorithm. Finally, we demonstrate the performance results for dynamic system identification via simulations. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Image compression based on wavelet transform and vector quantization

    Page(s): 1778 - 1780 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (277 KB) |  | HTML iconHTML  

    This paper presents an image compression scheme that uses the wavelet transform and neural network. Firstly, image is decomposed at different scales by using the wavelet transform. Then, the different quantization and. coding schemes for each sub-image are carried out in accordance with its statistical properties and distributed properties of the wavelet coefficients. The wavelet coefficients in low frequency subimage are. transformed by DCT and then they are compressed by using DPCM while the wavelet coefficients in high frequency sub-images are compressed and vector quantized by using Kohonen neural network on SOFM algorithm. Using these compressing techniques, we can obtain rather satisfactory reconstructed images with large compress ratio. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Research on controlling congestion in wireless mobile Internet via satellite based on multi-information and fuzzy identification technologies

    Page(s): 1697 - 1701 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (388 KB) |  | HTML iconHTML  

    With emerging problems of limited wireless spectrum resources and more and more wireless or mobile users in Internet, network congestion will be become an outstanding matter that needs be solved urgently in booming wireless mobile Internet. In order to meet the higher standard of wireless transmission in wireless mobile Internet via satellite for multimedia communication based on multi-information system with integrating speech, data, video and cartoon. A strategy for controlling congestion based on multi-information and fuzzy identification technologies is presented. The model of strategy for controlling congestion has a smart characteristic that used for analyzing and controlling congestion in wireless mobile Internet via satellite, moreover it possesses some basis functions for dynamically building wireless local area network oriented 3G and 4G and next generation Internet in the future communication networks and next generation network. Simulation shown: there are a certain self-adaptive control characters of congestion in the model of strategy for controlling congestion. However it is that maybe more developed to become an intelligence toolbox orient-large-scale network in building next generation Internet and next generation network. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A nonlinear combining forecast method based on fuzzy neural network

    Page(s): 2160 - 2164 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (398 KB) |  | HTML iconHTML  

    It has been shown in previous economic and statistical studies that combining forecasts may produce more accurate forecasts than individual ones. However, the literature on combining forecasts has almost exclusively focused on linear combined forecasts. The issues and methods of nonlinear combined forecasts have not yet been fully explored, even though forecast improvements may be possible using nonlinear combination techniques. We investigate the fuzzy neural network (FNN) as a tool for nonlinear combined forecasts. The performance of the networks is evaluated by comparing them to two individual forecasting methods and three conventional linear combining methods. The outcome of the comparison proved that the prediction by the FNN method generally performs better than those by individual forecasting methods, as well as linear combining methods. The paper suggests that the FNN method can be used as an alternative to conventional linear combining methods to achieve greater forecasting accuracy. Superiority of the FNN arises because of its flexibility in accounting for potentially complex nonlinear relationships not easily captured by traditional linear models. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Individual character font recognition based on guidance font

    Page(s): 1715 - 1717 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (291 KB) |  | HTML iconHTML  

    Font recognition of Chinese character is an important part of Chinese character recognition and page layout reconstruction. This article puts forward a new font recognition method. The guidance fonts are acquired from the chapter font or acquired knowledge of font typesetting, and it starts the corresponding single-font character recognizer to get the matching results that are used in recognizing fonts. Experiment shows that this method increases the speed and veracity of recognition. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Image retrieval using object template

    Page(s): 2090 - 2096 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (468 KB) |  | HTML iconHTML  

    How to mine, organize and use the information of the objects in an image, which can just reflect the user's intension precisely, is overlooked for a long time. From the point of view of mining, training and utilizing such information efficiently, this paper focuses on a particular group of features for every class of object. With this group of features, the class of object can be separated with others most distinctly. Based on this idea, using SVM (Support Vector Machine), HGM (Hybrid Gauss Model), PCA (principle component analysis) and FFC (Forward Feature Combination), we design and implement an Object Template that can represent a kind of object (such as tiger) perfectly and then present its two applications. With our Object Template, the Recall Precision (P100) for image retrieval can be increased to 46.175%; the precision of Object Detection also reaches 82.832%. (On 3400 real-world image collections). View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A wavelet-based method for fingerprint image enhancement

    Page(s): 1973 - 1977 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (620 KB) |  | HTML iconHTML  

    The performance of AFIS (automatic fingerprint identification system) is heavily determined by the quality of the input image. Thus an effective method to enhance the fingerprint image is essential in each system. A new method based on the wavelet decomposition is proposed in this paper. We combine the texture filter method mostly used in nowadays with wavelet transform to achieve a more reliable and effective enhancement. Using this method, we obtain a more clarity fingerprint image, which can distinctly improve the precision of minutiae extraction module. Experimental results show that wavelet-based enhancement method is more effective and robust than the other existing methods such as filter-based and direct grey level approaches. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The design of telediagnosis system model based on mobile agent

    Page(s): 1737 - 1741 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (357 KB) |  | HTML iconHTML  

    The paper deals with the telediagnosis system which applies a mobile agent mechanism to the medical server field. Through the discussion and comment on the key problems, an effective solution is derived for the system. First, to perfect the telediagnosis process, the system adopts the strategy of combining the centralized management with the distributed management. The diagnostic center (DC) controls the telediagnosis course; while the Sufferer Agent and Consultation Agent take responsibility for the concrete implementation. Next, the system uses the object redundance server technique and agent fault-tolerant technique to improve the reliability and throughput ratio. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Skeletonization of ribbon-like shapes with new wavelet function

    Page(s): 1869 - 1874 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (461 KB)  

    In this paper we propose a new scheme to extract the skeleton of ribbon-like shape with a novel wavelet function. It consists of two phases based on these perfect properties of the new wavelet function; and symmetry analyses of maxima moduli of wavelet transform are given. Midpoints of all pairs of contour elements are connected to generate a skeleton of the shape, which is defined as wavelet skeleton. Four basic criteria for modifying the artifacts of wavelet skeleton are presented. A corresponding algorithm is developed, and the experimental results are shown that this algorithm is capable of extracting exactly the skeleton of ribbon-like shape with different widths as well as different grey-levels. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Stock share analysis based on Gray system relevance theory

    Page(s): 1781 - 1783 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (255 KB) |  | HTML iconHTML  

    This paper points out that the developing stock market of China has now had the feature of behaving separately as a huge market, instead of the rising/dropping-all-together feature of the small-scale market. But it is still immature with the stockjobbing character of fund driven rise. To follow the major fund is an effective way to win solid and high reward in current stock market in China. A scientific method to select shares objectively and quantificationally is also derivate from the gray system relevance analysis. This method is scientific, maneuverable, and practicable. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • DCA-Tree: a high performance structure for incremental update cube on MDDW

    Page(s): 2069 - 2072 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (320 KB) |  | HTML iconHTML  

    MOLAP (multidimensional online analytical processing) is an important application on multidimensional data warehouse. We often execute range queries on aggregate cube computed by pre-aggregate technique in MOLAP. In this paper, we propose an high performance and incremental update cube which can reduce the update cost significantly while maintaining reasonable search efficiency, by using an index structure called the data cube aggregate tree (DCA-Tree) which is an improved version of the R*-Tree. The DCA-Tree stores the aggregate values for the minimum bounding rectangle (MBR) of the data cube by some pre-aggregate technology and thus to minimize the update cost since only a small fraction of data nodes in the data cube is changed. When we insert the new dimension, the DCA-Tree locates the address of the insertion node and inserts the new data nodes of the new insert dimension into the double linked list of the data nodes by modify the pointers of those data nodes. Then incrementally update the affected the ancestor's of the insertion data nodes from the data nodes to the root node. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Image recognition based on evolutionary algorithm

    Page(s): 1771 - 1773 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (308 KB) |  | HTML iconHTML  

    In this article, we propose two methods based on the evolutionary algorithm to recognize simple and complex patterns in complex images with a lot of noise information. We also adjust the fitness function in the evolutionary algorithm to speed up the computation. Several experiments have also been made to show that through these methods and adjustment, via comparisons with fixed patterns, objects with different X and Y direction scale factors, rotation angle and translation in complex images can be recognized easily with very quite high precision. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.