Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1199030
A. Galushkin, V. S. Zlobin, S. V. Korobkova, E.I. Rjabtsev, N. Tomashevich, E.P. Tumoian
The development of the system of recognition of the different contagions, including tuberculosis contageons, is conducted in the Scientific Center of Neurocomputers. In this work some algorithms of tuberculosis contageons (Koch's bacillus) identification are presented. They were designed in the Scientific Center of Neurocomputers RACS in the context of solving this task. The description of Express algorithm and the algorithms of neural network processing of the images and the results are adduced.
{"title":"Neurocomputer processing of the images in the task of tuberculosis contageons identification","authors":"A. Galushkin, V. S. Zlobin, S. V. Korobkova, E.I. Rjabtsev, N. Tomashevich, E.P. Tumoian","doi":"10.1109/ICONIP.2002.1199030","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1199030","url":null,"abstract":"The development of the system of recognition of the different contagions, including tuberculosis contageons, is conducted in the Scientific Center of Neurocomputers. In this work some algorithms of tuberculosis contageons (Koch's bacillus) identification are presented. They were designed in the Scientific Center of Neurocomputers RACS in the context of solving this task. The description of Express algorithm and the algorithms of neural network processing of the images and the results are adduced.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133992516","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}
Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1198194
Z.H. Zhang, C. Ong, S. Keerthi, E.G. Gilbert
The paper presents a new approach to determine the stability region for constrained dynamical systems. Our approach employs support vector machines (SVMs), a promising new tool for pattern recognition, to this field. By this application, the determination of stability region becomes a typical two-class hard margin pattern recognition problem, rather than the characterizations of the boundaries of such stability regions. In the underlying analysis, a program has been developed to generate critical points in the state space and train them by SVMs. Some examples are given to show the obtained estimates are close approximations of the exact stability region.
{"title":"Using support vector machines for stability region determination","authors":"Z.H. Zhang, C. Ong, S. Keerthi, E.G. Gilbert","doi":"10.1109/ICONIP.2002.1198194","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1198194","url":null,"abstract":"The paper presents a new approach to determine the stability region for constrained dynamical systems. Our approach employs support vector machines (SVMs), a promising new tool for pattern recognition, to this field. By this application, the determination of stability region becomes a typical two-class hard margin pattern recognition problem, rather than the characterizations of the boundaries of such stability regions. In the underlying analysis, a program has been developed to generate critical points in the state space and train them by SVMs. Some examples are given to show the obtained estimates are close approximations of the exact stability region.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131768933","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}
Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1198181
Jiaxiang Bi
We describe and discuss the coding method and self-organizing process of the information in the brain, advance a kind of encoding system: the "natural encoding system" which is particular to nature itself. In the problem of machine imitation, we further discuss the topological property equivalence problem of artificial and natural intelligence system on the basis of hierarchy structures of the virtual machine. More specifically we present the three large hierarchy structures of the intelligence system: the physical hierarchy, the physiological hierarchy and the psychological hierarchy. We also describe the relations between the three large hierarchies and some sub-stratums, the encoding method of information, the holographic frame structure of the information, the functions of the virtual machine system on different levels, etc. in more detail.
{"title":"Research about holographic relation and topological structure for thinking process on brain and artificial intelligence system","authors":"Jiaxiang Bi","doi":"10.1109/ICONIP.2002.1198181","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1198181","url":null,"abstract":"We describe and discuss the coding method and self-organizing process of the information in the brain, advance a kind of encoding system: the \"natural encoding system\" which is particular to nature itself. In the problem of machine imitation, we further discuss the topological property equivalence problem of artificial and natural intelligence system on the basis of hierarchy structures of the virtual machine. More specifically we present the three large hierarchy structures of the intelligence system: the physical hierarchy, the physiological hierarchy and the psychological hierarchy. We also describe the relations between the three large hierarchies and some sub-stratums, the encoding method of information, the holographic frame structure of the information, the functions of the virtual machine system on different levels, etc. in more detail.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134224878","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}
Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1198156
O. Hoshino, K. Mitsunaga, M. Miyamoto, K. Kuroiwa
By simulating a neural network model we investigated roles of background spectral components of vowel sounds in the neuronal representation of vowel sounds. The model consists of two networks, by which vowel sounds are processed in a hierarchical manner. The first network, which is tonotopically organized, detects spectral peaks called first and second formant frequencies (F1 and F2). The second network has a tonotopic two-dimensional structure and receives input from the first network in a convergent manner. The second network detects the combinatory information of the first (F1) and second (F2) formant frequencies of vowel sounds. We trained the model with five Japanese vowels spoken by different people and modified synaptic connection strengths of the second network according to the Hebbian learning rule, by which relevant dynamic cell assemblies expressing categories of vowels were organized. We show that for creating the dynamic cell assemblies background components around two-formant peaks (F1, F2) are not necessary but advantageous for the creation of the cell assemblies.
{"title":"Dynamic cell assemblies and vowel sound categorization","authors":"O. Hoshino, K. Mitsunaga, M. Miyamoto, K. Kuroiwa","doi":"10.1109/ICONIP.2002.1198156","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1198156","url":null,"abstract":"By simulating a neural network model we investigated roles of background spectral components of vowel sounds in the neuronal representation of vowel sounds. The model consists of two networks, by which vowel sounds are processed in a hierarchical manner. The first network, which is tonotopically organized, detects spectral peaks called first and second formant frequencies (F1 and F2). The second network has a tonotopic two-dimensional structure and receives input from the first network in a convergent manner. The second network detects the combinatory information of the first (F1) and second (F2) formant frequencies of vowel sounds. We trained the model with five Japanese vowels spoken by different people and modified synaptic connection strengths of the second network according to the Hebbian learning rule, by which relevant dynamic cell assemblies expressing categories of vowels were organized. We show that for creating the dynamic cell assemblies background components around two-formant peaks (F1, F2) are not necessary but advantageous for the creation of the cell assemblies.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"61 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130332703","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}
Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1198952
K. Kitano, T. Fukai
The intrinsic properties of 'synfire chain', the feedforward network propagating synchronous spike packets, has been studied so far. Possible functional roles of the synfire chain, however, has been poorly understood. Considering that coordinated activities of multiple synfire chains can serve as a reference time, we study whether a network model based on the multiple synfire chains contributes to generation of predictive synchrony to occurrence times of external events, observed in the primary motor cortex. In our model, neurons that code occurrence times of external events are partly innervated by the multiple synfire chains. The event times are embedded into the synaptic projections between layers that coincide with the events and event coding neurons through spike-timing-dependent synaptic learning. From our simulation results, it is found that our model can generate the predictive synchrony when the ratio of the projections is within a suitable range.
{"title":"A multiple synfire-chain model for the predictive synchrony in the motor-related cortical areas","authors":"K. Kitano, T. Fukai","doi":"10.1109/ICONIP.2002.1198952","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1198952","url":null,"abstract":"The intrinsic properties of 'synfire chain', the feedforward network propagating synchronous spike packets, has been studied so far. Possible functional roles of the synfire chain, however, has been poorly understood. Considering that coordinated activities of multiple synfire chains can serve as a reference time, we study whether a network model based on the multiple synfire chains contributes to generation of predictive synchrony to occurrence times of external events, observed in the primary motor cortex. In our model, neurons that code occurrence times of external events are partly innervated by the multiple synfire chains. The event times are embedded into the synaptic projections between layers that coincide with the events and event coding neurons through spike-timing-dependent synaptic learning. From our simulation results, it is found that our model can generate the predictive synchrony when the ratio of the projections is within a suitable range.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130382198","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}
Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1202797
J. Torreão
Binocular disparities arise from the positional differences of scene features projected in the two retinae. Disparity-selective neurons are known to exist in several areas of the visual cortex of cats and monkeys, and have been associated with mechanisms of gaze stabilization and stereoscopic depth perception. Such neurons appear with different response profiles, leading to their classification as tuned excitatory, tuned inhibitory, tuned near, tuned far, and reciprocal (near and far) neurons. Here we propose an analytical model for the shape of these disparity selectivity curves, showing that they can be approximated as either the Green's function or the homogeneous solution to a second-order differential equation derived from a signal matching constraint. This means that the mathematical solution to the matching problem involves functions which are similar in shape to the selectivity profiles of the binocular neurons.
{"title":"An analytical model for the disparity selectivity profiles of binocular neurons","authors":"J. Torreão","doi":"10.1109/ICONIP.2002.1202797","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1202797","url":null,"abstract":"Binocular disparities arise from the positional differences of scene features projected in the two retinae. Disparity-selective neurons are known to exist in several areas of the visual cortex of cats and monkeys, and have been associated with mechanisms of gaze stabilization and stereoscopic depth perception. Such neurons appear with different response profiles, leading to their classification as tuned excitatory, tuned inhibitory, tuned near, tuned far, and reciprocal (near and far) neurons. Here we propose an analytical model for the shape of these disparity selectivity curves, showing that they can be approximated as either the Green's function or the homogeneous solution to a second-order differential equation derived from a signal matching constraint. This means that the mathematical solution to the matching problem involves functions which are similar in shape to the selectivity profiles of the binocular neurons.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114381856","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}
Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1199008
J. Dávila
This paper presents a method for analyzing GA effectiveness for the evolution of neural networks. The analysis is based on the schemata of the (phenotype) neural network being evolved, as opposed to the traditional method of analyzing schemata disruptions at the genotype level. Comparisons between the two types of analysis are made. Empirical data is presented that indicates the greater validity of the analysis at the phenotype level.
{"title":"Disruption analysis for neural network topology evolution systems","authors":"J. Dávila","doi":"10.1109/ICONIP.2002.1199008","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1199008","url":null,"abstract":"This paper presents a method for analyzing GA effectiveness for the evolution of neural networks. The analysis is based on the schemata of the (phenotype) neural network being evolved, as opposed to the traditional method of analyzing schemata disruptions at the genotype level. Comparisons between the two types of analysis are made. Empirical data is presented that indicates the greater validity of the analysis at the phenotype level.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"109 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124670840","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}
Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1198219
M. Palaniswami, A. Shilton
Support vector machines are a general formulation for machine learning. It has been shown to perform extremely well for a number of problems in classification and regression. However, in many difficult problems, the system dynamics may change with time and the resulting new information arriving incrementally will provide additional data. At present, there is limited work to cope with the computational demands of modeling time varying systems. Therefore, we develop the concept of adaptive support vector machines that can learn from incremental data. Results are provided to demonstrate the applicability of the adaptive support vector machines techniques for pattern classification and regression problems.
{"title":"Adaptive support vector machines for regression","authors":"M. Palaniswami, A. Shilton","doi":"10.1109/ICONIP.2002.1198219","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1198219","url":null,"abstract":"Support vector machines are a general formulation for machine learning. It has been shown to perform extremely well for a number of problems in classification and regression. However, in many difficult problems, the system dynamics may change with time and the resulting new information arriving incrementally will provide additional data. At present, there is limited work to cope with the computational demands of modeling time varying systems. Therefore, we develop the concept of adaptive support vector machines that can learn from incremental data. Results are provided to demonstrate the applicability of the adaptive support vector machines techniques for pattern classification and regression problems.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124754538","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}
Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1202172
S. K. Manandhar, R. Sadananda
Although the Hopfield network can store and retrieve patterns, its storage capacity is limited. In this study we investigate the effect of Hamming distance of stored patterns on the success of their retrieval. The results show that by removing patterns having low Hamming distance with each other, the capacity of the network increases.
{"title":"Effect of Hamming distance of patterns on storage capacity of Hopfield network","authors":"S. K. Manandhar, R. Sadananda","doi":"10.1109/ICONIP.2002.1202172","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1202172","url":null,"abstract":"Although the Hopfield network can store and retrieve patterns, its storage capacity is limited. In this study we investigate the effect of Hamming distance of stored patterns on the success of their retrieval. The results show that by removing patterns having low Hamming distance with each other, the capacity of the network increases.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132530053","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}
Pub Date : 2002-11-18DOI: 10.1109/ICONIP.2002.1198214
Xiuju Fu, Lipo Wang, K. Chua, Feng Chu
This paper presents a new algorithm for the construction and training of an RBF neural network with unbalanced data. In applications, minority classes with much fewer samples are often present in data sets. The learning process of a neural network usually is biased towards classes with majority populations. Our study focused on improving the classification accuracy of minority classes while maintaining the overall classification performance.
{"title":"Training RBF neural networks on unbalanced data","authors":"Xiuju Fu, Lipo Wang, K. Chua, Feng Chu","doi":"10.1109/ICONIP.2002.1198214","DOIUrl":"https://doi.org/10.1109/ICONIP.2002.1198214","url":null,"abstract":"This paper presents a new algorithm for the construction and training of an RBF neural network with unbalanced data. In applications, minority classes with much fewer samples are often present in data sets. The learning process of a neural network usually is biased towards classes with majority populations. Our study focused on improving the classification accuracy of minority classes while maintaining the overall classification performance.","PeriodicalId":146553,"journal":{"name":"Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02.","volume":"127 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131943721","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：