Pub Date : 2002-08-07DOI: 10.1109/IJCNN.2002.1007595
W. W. Hsieh, Aiming Wu
Singular spectrum analysis (SSA), a linear univariate and multivariate time series technique, is essentially principal component analysis (PCA) applied to the time series and additional copies of the time series lagged by 1 to L-1 time steps. Neural network theory has meanwhile allowed PCA to be generalized to nonlinear PCA (NLPCA). In the paper, NLPCA is further extended to perform nonlinear SSA (NLSSA). First, SSA is applied to the data, then the leading principal components of the SSA are chosen as inputs to an NLPCA network (with a circular node at the bottleneck), which performs the NLSSA by nonlinearly combining all the input SSA modes into a single NLSSA mode. This nonlinear spectral technique allows the detection of highly anharmonic oscillations, as illustrated by a stretched square wave imbedded in white noise, which shows NLSSA to be superior to SSA and classical Fourier spectral analysis.
{"title":"Nonlinear singular spectrum analysis","authors":"W. W. Hsieh, Aiming Wu","doi":"10.1109/IJCNN.2002.1007595","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1007595","url":null,"abstract":"Singular spectrum analysis (SSA), a linear univariate and multivariate time series technique, is essentially principal component analysis (PCA) applied to the time series and additional copies of the time series lagged by 1 to L-1 time steps. Neural network theory has meanwhile allowed PCA to be generalized to nonlinear PCA (NLPCA). In the paper, NLPCA is further extended to perform nonlinear SSA (NLSSA). First, SSA is applied to the data, then the leading principal components of the SSA are chosen as inputs to an NLPCA network (with a circular node at the bottleneck), which performs the NLSSA by nonlinearly combining all the input SSA modes into a single NLSSA mode. This nonlinear spectral technique allows the detection of highly anharmonic oscillations, as illustrated by a stretched square wave imbedded in white noise, which shows NLSSA to be superior to SSA and classical Fourier spectral analysis.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123567729","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 : 2002-08-07DOI: 10.1109/IJCNN.2002.1007445
A. J. Anderson, P. McOwan
A sensorimotor controller for a biologically inspired stealth strategy (motion camouflage) is implemented in a software simulation using backpropagation. When operating with realistic inputs, the controller allows a predator to track a prey that moves along real hoverfly flight paths, whilst appearing to remain stationary.
{"title":"Towards an autonomous motion camouflage control system","authors":"A. J. Anderson, P. McOwan","doi":"10.1109/IJCNN.2002.1007445","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1007445","url":null,"abstract":"A sensorimotor controller for a biologically inspired stealth strategy (motion camouflage) is implemented in a software simulation using backpropagation. When operating with realistic inputs, the controller allows a predator to track a prey that moves along real hoverfly flight paths, whilst appearing to remain stationary.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126626016","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 : 2002-08-07DOI: 10.1109/IJCNN.2002.1007658
I. Nunes de Silva, L. Nepomuceno, T. M. Bastos
Economic dispatch (ED) problems have recently been solved by artificial neural networks approaches. In most of these dispatch models, the cost function must be linear or quadratic. Therefore, functions that have several minimum points represent a problem to the simulation since these approaches have not accepted nonlinear cost function. Another drawback pointed out in the literature is that some of these neural approaches fail to converge efficiently towards feasible equilibrium points. This paper discusses the application of a modified Hopfield architecture for solving ED problems defined by nonlinear cost function. The internal parameters of the neural network adopted here are computed using the valid-subspace technique, which guarantees convergence to equilibrium points that represent a solution for the ED problem. Simulation results and a comparative analysis involving a 3-bus test system are presented to illustrate efficiency of the proposed approach.
{"title":"Designing a modified Hopfield network to solve an economic dispatch problem with nonlinear cost function","authors":"I. Nunes de Silva, L. Nepomuceno, T. M. Bastos","doi":"10.1109/IJCNN.2002.1007658","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1007658","url":null,"abstract":"Economic dispatch (ED) problems have recently been solved by artificial neural networks approaches. In most of these dispatch models, the cost function must be linear or quadratic. Therefore, functions that have several minimum points represent a problem to the simulation since these approaches have not accepted nonlinear cost function. Another drawback pointed out in the literature is that some of these neural approaches fail to converge efficiently towards feasible equilibrium points. This paper discusses the application of a modified Hopfield architecture for solving ED problems defined by nonlinear cost function. The internal parameters of the neural network adopted here are computed using the valid-subspace technique, which guarantees convergence to equilibrium points that represent a solution for the ED problem. Simulation results and a comparative analysis involving a 3-bus test system are presented to illustrate efficiency of the proposed approach.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"44 1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115551679","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 : 2002-08-07DOI: 10.1109/IJCNN.2002.1007653
K. Aquere, J. Quillfeldt, R. D. de Almeida
In this work we consider a neural network model with spatially and temporally structured synapses whose dynamics may depend on more than one time step. This model is capable of storing and recovering temporal sequences or cycles. Hebb-like learning rules are used to store the temporal sequences of patterns and Hamming-like distance for cycles is defined to measure the distance between two different cycles. We perform a signal-to-noise analysis of the system and numerically determine the critical capacity of the network, basins of attractions size, stability of recovery states and investigate the effects of spurious states in the performance of the net. We show that the performance of the net is enhanced when information is stored in temporally longer sequences.
在这项工作中，我们考虑了一个具有空间和时间结构突触的神经网络模型，其动力学可能依赖于多个时间步长。该模型能够存储和恢复时间序列或周期。Hebb-like学习规则用于存储模式的时间序列，hming -like distance for cycles用于度量两个不同循环之间的距离。我们对系统进行了信噪分析，并在数值上确定了网络的临界容量、吸引力盆地的大小、恢复状态的稳定性，并研究了虚假状态对网络性能的影响。我们表明，当信息存储在时间较长的序列中时，网络的性能得到增强。
{"title":"Load capacity of a neural network model with spatially and temporally structured connectivity","authors":"K. Aquere, J. Quillfeldt, R. D. de Almeida","doi":"10.1109/IJCNN.2002.1007653","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1007653","url":null,"abstract":"In this work we consider a neural network model with spatially and temporally structured synapses whose dynamics may depend on more than one time step. This model is capable of storing and recovering temporal sequences or cycles. Hebb-like learning rules are used to store the temporal sequences of patterns and Hamming-like distance for cycles is defined to measure the distance between two different cycles. We perform a signal-to-noise analysis of the system and numerically determine the critical capacity of the network, basins of attractions size, stability of recovery states and investigate the effects of spurious states in the performance of the net. We show that the performance of the net is enhanced when information is stored in temporally longer sequences.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"363 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122770652","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 : 2002-08-07DOI: 10.1109/IJCNN.2002.1007747
K. Ahmad, B. Vrusias, M. Tariq
'Integrated' classification refers to the conjunctive or competitive use of two or more (neural) classifiers. A cooperative neural network system comprising two independently trained Kohonen networks and co-operating with the help of a Hebbian network, is described. The effectiveness of such a network is demonstrated by using it to retrieve images and related texts from a multi-media database. Preliminary results of such an approach appear to be encouraging.
{"title":"Co-operative neural networks and 'integrated' classification","authors":"K. Ahmad, B. Vrusias, M. Tariq","doi":"10.1109/IJCNN.2002.1007747","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1007747","url":null,"abstract":"'Integrated' classification refers to the conjunctive or competitive use of two or more (neural) classifiers. A cooperative neural network system comprising two independently trained Kohonen networks and co-operating with the help of a Hebbian network, is described. The effectiveness of such a network is demonstrated by using it to retrieve images and related texts from a multi-media database. Preliminary results of such an approach appear to be encouraging.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114256003","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 : 2002-08-07DOI: 10.1109/IJCNN.2002.1007818
A. Paasio, M. Laiho, A. Kananen, K. Halonen
This paper describes functionalities which will soon be available in an analog array processor. By developing dedicated calculation cores for different types of applications, the processor size can be kept small and therefore the achieved resolution is relatively high. The available features include weighted ranked order filtering, gray scale mathematical morphology, second neighbor direct interaction and different plasticity options for templates, where the previous results of the processing define the weights in the future steps. A linear filter capable of low pass filtering is also included. The resulting processing unit is viewed at block level and the characteristics of functional blocks are assessed in terms of estimations of power consumption, evaluation time, die area and accuracy.
{"title":"An analog array processor hardware realization with multiple new features","authors":"A. Paasio, M. Laiho, A. Kananen, K. Halonen","doi":"10.1109/IJCNN.2002.1007818","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1007818","url":null,"abstract":"This paper describes functionalities which will soon be available in an analog array processor. By developing dedicated calculation cores for different types of applications, the processor size can be kept small and therefore the achieved resolution is relatively high. The available features include weighted ranked order filtering, gray scale mathematical morphology, second neighbor direct interaction and different plasticity options for templates, where the previous results of the processing define the weights in the future steps. A linear filter capable of low pass filtering is also included. The resulting processing unit is viewed at block level and the characteristics of functional blocks are assessed in terms of estimations of power consumption, evaluation time, die area and accuracy.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"215 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114399709","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 : 2002-08-07DOI: 10.1109/IJCNN.2002.1005534
V. Gosasang, T. Tanprasert
This paper presents an analysis on the integration of Neural Network hardware to PC and a solution to the cache coherence problem. An analysis is achieved by determining clock cycles in CPU operation compared to mixed CPU-ANN mode. Cache coherence problem is resolved by hardware-based protocol executed on an additional cache consistency controller.
{"title":"Performance and caching issues in an integration of neural net and conventional PC","authors":"V. Gosasang, T. Tanprasert","doi":"10.1109/IJCNN.2002.1005534","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1005534","url":null,"abstract":"This paper presents an analysis on the integration of Neural Network hardware to PC and a solution to the cache coherence problem. An analysis is achieved by determining clock cycles in CPU operation compared to mixed CPU-ANN mode. Cache coherence problem is resolved by hardware-based protocol executed on an additional cache consistency controller.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"57 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117243794","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 : 2002-08-07DOI: 10.1109/IJCNN.2002.1007784
S. Chiewchanwattana, C. Lursinsap
This paper considers the problem of incomplete time-series prediction by FI-GEM (fill-in-generalized ensemble method) networks, which has two steps. The first step is composed of several fill-in methods for preprocessing the missing value of time-series and the outcome are the complete time-series data. The second step is composed of the several individual multilayer perceptrons (MLP) whose their outputs are combined by the generalized ensemble method. There are five fill-in methods that are explored: cubic smoothing spline interpolation, and four imputation methods: EM (expectation maximization), regularized EM, average EM, average regularized EM. Mackey-Glass chaotic time-series and sunspots data are used for evaluating our approach. The experimental results show that the prediction accuracy of FI-GEM networks are much better than individual neural networks.
{"title":"FI-GEM networks for incomplete time-series prediction","authors":"S. Chiewchanwattana, C. Lursinsap","doi":"10.1109/IJCNN.2002.1007784","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1007784","url":null,"abstract":"This paper considers the problem of incomplete time-series prediction by FI-GEM (fill-in-generalized ensemble method) networks, which has two steps. The first step is composed of several fill-in methods for preprocessing the missing value of time-series and the outcome are the complete time-series data. The second step is composed of the several individual multilayer perceptrons (MLP) whose their outputs are combined by the generalized ensemble method. There are five fill-in methods that are explored: cubic smoothing spline interpolation, and four imputation methods: EM (expectation maximization), regularized EM, average EM, average regularized EM. Mackey-Glass chaotic time-series and sunspots data are used for evaluating our approach. The experimental results show that the prediction accuracy of FI-GEM networks are much better than individual neural networks.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129745482","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 : 2002-08-07DOI: 10.1109/IJCNN.2002.1005586
A. Meyer-Baese
The dynamics of complex neural networks must include the aspects of long and short-term memory. The behaviour of the network is characterized by an equation of neural activity as a fast phenomenon and an equation of synaptic modification as a slow part of the neural system. We present a method of analyzing the dynamics of a system with different time scales based on the theory of flow invariance. We are able to show the conditions under which the solutions of such a system are bounded being less restrictive than with the K-monotone theory.
{"title":"Flow invariance for competitive neural networks with different time-scales","authors":"A. Meyer-Baese","doi":"10.1109/IJCNN.2002.1005586","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1005586","url":null,"abstract":"The dynamics of complex neural networks must include the aspects of long and short-term memory. The behaviour of the network is characterized by an equation of neural activity as a fast phenomenon and an equation of synaptic modification as a slow part of the neural system. We present a method of analyzing the dynamics of a system with different time scales based on the theory of flow invariance. We are able to show the conditions under which the solutions of such a system are bounded being less restrictive than with the K-monotone theory.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129845540","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 : 2002-08-07DOI: 10.1109/IJCNN.2002.1007689
T. Asai, Y. Amemiya
We present an inhibitory neural network implemented on analog CMOS chips, whose neurons compete with each other in the frequency and time domains. The circuit for each neuron was designed to produce sequences in time of identically shaped pulses, called spikes. The results of experiments and simulations revealed that the network more efficiently achieved the selective activation and inactivation of the neural circuits on the basis of spike timing than on the basis of firing rates. The results indicate that neural processing based on the spike timing of neural circuits provides a possible way to overcome the low-tolerance problems of analog devices in noisy environments.
{"title":"Frequency- and temporal-domain neural competition in analog integrate-and-fire neurochips","authors":"T. Asai, Y. Amemiya","doi":"10.1109/IJCNN.2002.1007689","DOIUrl":"https://doi.org/10.1109/IJCNN.2002.1007689","url":null,"abstract":"We present an inhibitory neural network implemented on analog CMOS chips, whose neurons compete with each other in the frequency and time domains. The circuit for each neuron was designed to produce sequences in time of identically shaped pulses, called spikes. The results of experiments and simulations revealed that the network more efficiently achieved the selective activation and inactivation of the neural circuits on the basis of spike timing than on the basis of firing rates. The results indicate that neural processing based on the spike timing of neural circuits provides a possible way to overcome the low-tolerance problems of analog devices in noisy environments.","PeriodicalId":382771,"journal":{"name":"Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2002-08-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128205654","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：