L. Zhong, Jingling Yuan, Hongxia Xia, Chengming Zou
{"title":"A study on gray neural network modeling","authors":"L. Zhong, Jingling Yuan, Hongxia Xia, Chengming Zou","doi":"10.1109/ICMLC.2002.1175391","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1175391","url":null,"abstract":"In order to resolve complex gray uncertainly problems, the gray neural network modeling (GNNM) method by combining gray system theory with neural network is studied and improved in the paper. Also, a method of GNNM (1, 1) based on the time response model is proposed. With an eye to the multidimensional indefinite problems in which all kinds of genetic relationships and mechanisms exist, GNNM (1, 4) is established on the basis of GNNM (1, 1).","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"55 1","pages":"2021-2023 vol.4"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84835802","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"The study of epistasis based on the random walk model in fitness landscapes of schemata","authors":"Jian-Wu Li, Min-Qiang Li","doi":"10.1109/ICMLC.2002.1167435","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1167435","url":null,"abstract":"Interdependence between genes is an important factor causing hardness in genetic algorithms (GA). Traditional methods, which are used to measure the interaction between genes, can only reflect the extent of epistasis between all genes in the chromosome. In this paper, we propose the definition of the fitness landscape of schemata, and perform random walks on this landscape to study the degree of interdependence between some certain gene loci in study. According to the degree of interaction between these given gene loci, we can analyze and determine building blocks of GA. We also do a lot of experiments based on NK-models, and results of empirical analysis show that this method is effective.","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"107 1","pages":"1396-1400 vol.3"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77595437","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Matrix criterion for dynamic analysis in discrete neural networks with multiple delays","authors":"Eric C. C. Tsang, S. Qiu, D. Yeung","doi":"10.1109/ICMLC.2002.1175439","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1175439","url":null,"abstract":"The dynamics of a discrete Hopfield neural network with multiple delays (HNNMDs) is studied by using a matrix inequality which is shown to be equivalent to the state transition equation of the HNNMDs network. Earlier work (2000) on discrete Hopfield neural networks showed that a parallel or serial mode of operation always leads to a limit cycle of period one or two for a skew or symmetric matrix, but they did not give an arbitrary weight matrix on how an updating operation might be needed to reach such a cycle. In this paper we present the existence conditions of limit cycles using matrix criteria in the HNNMDs network. For a network with an arbitrary weight matrix, the necessary and sufficient conditions for the existence of a limit cycle of period 1 and r are provided. The conditions for the existence of a special limit cycle of period 1 and 2 are also found. These results provide the foundation for many applications. A HNNMDs is said to have no stable state (fixed point) if it has a limit cycle of period 2 or more, which is stated in Theorem 5. A computer simulation demonstrates that the theoretical analysis in Theorem 5 is correct.","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"44 1","pages":"2245-2250 vol.4"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86906667","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Correlation based generating rules for fuzzy classification","authors":"Da-Zhong Liu, Xizhao Wang, J.W.T. Lee","doi":"10.1109/ICMLC.2002.1175333","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1175333","url":null,"abstract":"Automated acquisition of rules is a very important issue in knowledge engineering. In this paper, a new method for generating fuzzy rules from numerical data for fuzzy classification problems based on the correlation analysis between two fuzzy sets has been proposed. Experimental results and comparison with Yuan and Shaw's method are presented as well.","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"5 1","pages":"1733-1736 vol.4"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"87752930","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Li-xin Zhang, Yannan Zhao, Zehong Yang, Jiaxin Wang
{"title":"Feature selection in recognition of handwritten Chinese characters","authors":"Li-xin Zhang, Yannan Zhao, Zehong Yang, Jiaxin Wang","doi":"10.1109/ICMLC.2002.1167382","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1167382","url":null,"abstract":"Recognition of handwritten Chinese characters is a large-scale pattern recognition task, which is difficult and time consuming to build the corresponding classifiers. In this paper, two feature selection methods are proposed to reduce the complexity and speed up the handwritten Chinese recognition: one is the ReliefF-Wrapper method which evaluates the original features with the ReliefF method, and then uses the wrapper method to decide the number of features to be selected; and the other is GA-Wrapper that uses genetic algorithm to search the optimal subset of features with high training accuracy. Experiments were performed on 800 most frequently used Chinese characters, with 80,000 handwritten samples. Results show that the ReliefF-Wrapper method has good interpretation and high speed and GA-Wrapper gains higher accuracy. Limitations of the both methods and future work are also discussed.","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"20 1","pages":"1158-1162 vol.3"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90362531","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Combining multiple classifiers based on statistical method for handwritten Chinese character recognition","authors":"Lei Lin, Xiaolong Wang, Bingquan Liu","doi":"10.1109/ICMLC.2002.1176750","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1176750","url":null,"abstract":"In various application areas of pattern recognition, combining multiple classifiers is regarded as a method for achieving a substantial gain in performance of systems. The paper presents a method for handwritten Chinese character recognition to combine multiple classifiers based on statistics. Fusion strategies are discussed for providing a basis for combining classifiers. These combination strategies are experimentally tested on an online handwritten Chinese character recognition system. In our experiments, other combination approaches are also involved for comparison.","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"82 1","pages":"252-255 vol.1"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85972626","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Theoretical study on diversity of population in parallel genetic algorithms","authors":"Mei-Qin Pan, Guo-ping He","doi":"10.1109/ICMLC.2002.1176799","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1176799","url":null,"abstract":"In this paper, conditional probability density and marginal distribution are proposed as measures of population in genetic algorithms. The influence of selection, crossover and mutation on population distribution is analyzed. In addition, the recursive equations governing population density are derived, and a conclusion of global convergence is also shown.","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"2014 1","pages":"472-475 vol.1"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"86524913","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Medical image series segmentation using watershed transform and active contour model","authors":"F. Zhu, Jie Tian, Xiping Luo, Xingfei Ge","doi":"10.1109/ICMLC.2002.1174506","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1174506","url":null,"abstract":"In this paper, a semiautomatic algorithm based on the combination of the live wire algorithm and the active contour model is proposed for the segmentation of medical image series. First we obtain accurate segmentation of one or more slices in a medical image series by combining the livewire algorithm with the watershed method. Then the computer will segment the nearby slice using the modified active contour model. We introduce a gray-scale model to the boundary points of the active contour model to record the local region characters of the desired object in the segmented slice and replace the external energy of the traditional active contour model with the energy decided by the likelihood of the grayscale model. Moreover we introduce the active region concept of the snake to improve the segmentation accuracy. Experiment shows. that our algorithm can obtain the boundary of the desired object from a series of medical images reliably with only little user intervention.","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"23 1","pages":"865-870 vol.2"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81235336","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Research and implementation of a scalable secure active network node","authors":"Jianguo Wang, Zeng-zhi Li, Ya-nan Kou","doi":"10.1109/ICMLC.2002.1176720","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1176720","url":null,"abstract":"Active networks are packet-switched networks. It allows users to inject customized programs into the intermediate node of the network. The important difference of active network from traditional network is that active network consists of a series of active network nodes that are able to execute active codes and compute them. So, an active network node is the key of the architecture of active networks. This paper presents a scalable and secure active network node that provides the required flexibility of active network technology for dynamic services customized by users deployment and application-specific data processing and forwarding. In this paper, the composing components and services request interface and active application API are introduced, which make the active network node scalable, and the functions of the composing components can be dynamically extended. By describing the process procedure of capsules, we discuss the communication among these composing components. We develop a high performance active network node with a new mechanism to load an active code. We make the active network node secure by using a digital signature and secure execution engine that can authenticate the identities of principals and authorize to control resources accessing.","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"45 1","pages":"111-115 vol.1"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81331993","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"The approach of extracting features from the local environment for mobile robot","authors":"Wei Hong, Yan-tao Tian, Zai-li Dong","doi":"10.1109/ICMLC.2002.1174408","DOIUrl":"https://doi.org/10.1109/ICMLC.2002.1174408","url":null,"abstract":"A new data fusion method to extract features from the local environment for a mobile robot's navigation has been developed and implemented. This method, named the obstacle group, compresses data in a series of levels in order to reduce the quantity of data for communication between modules in a distributed single-robot system, or between all the robots and the central station in a multi-robot system. The method based on a grid map and an active window has strong adaptability and is real-time in a crowded environment. Experimental results demonstrate that the robot can successfully avoid collisions and plan its path by using this method.","PeriodicalId":90702,"journal":{"name":"Proceedings. International Conference on Machine Learning and Cybernetics","volume":"69 1","pages":"611-616 vol.2"},"PeriodicalIF":0.0,"publicationDate":"2002-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81441902","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}