Pub Date : 1991-08-15DOI: 10.1109/ICNN.1991.163360
R. Pridham, D.J. Hamilton
The problem of sonar signal discrimination of passive sonar events is addressed. Three generic systems are considered. The first is a conventional system that uses a quadratic Bayesian (QB) classifier. Next is a hybrid approach that uses a neural compound classifier network (CCN) of the type proposed by B.G. Batchelor (1974). Both the conventional and hybrid approaches use a generic automatic detector given by J.J. Wolcin (1984), which is structured to detect signals of arbitrary duration and frequency content. The third system is an all neural network approach which considers neural alternatives to the functions of detection, feature extraction, and feature optimization. The authors discuss a comparison of the first two systems. The third system is addressed by D.W. Cottle and D.J. Hamilton (ibid., this conference, p.13-19, 1991).<>
{"title":"Evaluation of neural network and conventional techniques for sonar signal discrimination","authors":"R. Pridham, D.J. Hamilton","doi":"10.1109/ICNN.1991.163360","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163360","url":null,"abstract":"The problem of sonar signal discrimination of passive sonar events is addressed. Three generic systems are considered. The first is a conventional system that uses a quadratic Bayesian (QB) classifier. Next is a hybrid approach that uses a neural compound classifier network (CCN) of the type proposed by B.G. Batchelor (1974). Both the conventional and hybrid approaches use a generic automatic detector given by J.J. Wolcin (1984), which is structured to detect signals of arbitrary duration and frequency content. The third system is an all neural network approach which considers neural alternatives to the functions of detection, feature extraction, and feature optimization. The authors discuss a comparison of the first two systems. The third system is addressed by D.W. Cottle and D.J. Hamilton (ibid., this conference, p.13-19, 1991).<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"78 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115070554","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 : 1991-08-15DOI: 10.1109/ICNN.1991.163355
A. Caiti, T. Parisini
A method for interpolating sparse measurements of ocean sediment properties by means of a network of parallel computational units is proposed. The network is able to generate a continuous mapping. of sediment properties as a function of x-y-z position, where z is the depth in the sediment, using a generalized radial basis function expansion. Advantages and disadvantages of the method are discussed, both from a physical and a computational viewpoint. An example with sediment density data obtained from sparse core measurements in a region of the Mediterranean sea is presented.<>
{"title":"Mapping of ocean sediments, by networks of parallel interpolating units","authors":"A. Caiti, T. Parisini","doi":"10.1109/ICNN.1991.163355","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163355","url":null,"abstract":"A method for interpolating sparse measurements of ocean sediment properties by means of a network of parallel computational units is proposed. The network is able to generate a continuous mapping. of sediment properties as a function of x-y-z position, where z is the depth in the sediment, using a generalized radial basis function expansion. Advantages and disadvantages of the method are discussed, both from a physical and a computational viewpoint. An example with sediment density data obtained from sparse core measurements in a region of the Mediterranean sea is presented.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"50 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123838936","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 : 1991-08-15DOI: 10.1109/ICNN.1991.163374
P. G. McKee, José M. F. Moura
The authors present a set of extensive experiments with alternative neural network, learning algorithms. These neural network configurations were tested on the problem of discriminating signals generated by an autoregressive moving-average (ARMA) linear system driven by white noise. These ARMA signals model a wide variety of signals arising in the ocean environment. The authors tested the various network models for their classification accuracy and speed of learning. The models investigated were back propagation, quickprop, Gaussian node networks, radial basis functions, the modified Kanerva methods, and networks without hidden units. For comparison, nearest-neighbor classifiers were also tested. Classification performance and learning time results are presented.<>
{"title":"Neural networks for classification of ARMA models: an experimental study","authors":"P. G. McKee, José M. F. Moura","doi":"10.1109/ICNN.1991.163374","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163374","url":null,"abstract":"The authors present a set of extensive experiments with alternative neural network, learning algorithms. These neural network configurations were tested on the problem of discriminating signals generated by an autoregressive moving-average (ARMA) linear system driven by white noise. These ARMA signals model a wide variety of signals arising in the ocean environment. The authors tested the various network models for their classification accuracy and speed of learning. The models investigated were back propagation, quickprop, Gaussian node networks, radial basis functions, the modified Kanerva methods, and networks without hidden units. For comparison, nearest-neighbor classifiers were also tested. Classification performance and learning time results are presented.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130890614","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}
The authors present a unique viewpoint in describing sea clutter. They demonstrate that the random nature of sea clutter is the result of chaotic phenomena. Using real-life sea clutter data, the authors use correlation dimension analysis to show that sea clutter can be embedded as a chaotic attractor in a finite-dimensional space. This observation provides a reliable indication for the existence of chaotic behavior. A neural network model incorporating the result of correlation-dimension analysis is used in the reconstruction of the dynamics of sea clutter. The model is in the form of a radial basis function network. The deterministic model for sea clutter is shown to be capable of predicting the evolution of sea clutter as a function of time.<>
{"title":"Neural network modeling of radar backscatter from an ocean surface using chaos theory","authors":"S. Haykin, H. Leung","doi":"10.1117/12.49784","DOIUrl":"https://doi.org/10.1117/12.49784","url":null,"abstract":"The authors present a unique viewpoint in describing sea clutter. They demonstrate that the random nature of sea clutter is the result of chaotic phenomena. Using real-life sea clutter data, the authors use correlation dimension analysis to show that sea clutter can be embedded as a chaotic attractor in a finite-dimensional space. This observation provides a reliable indication for the existence of chaotic behavior. A neural network model incorporating the result of correlation-dimension analysis is used in the reconstruction of the dynamics of sea clutter. The model is in the form of a radial basis function network. The deterministic model for sea clutter is shown to be capable of predicting the evolution of sea clutter as a function of time.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131174087","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 : 1991-08-15DOI: 10.1109/ICNN.1991.163352
P. Patrick, N. Ramani, W.G. Hanson, H. Anderson
The authors explore the potential of a neural network based system for detecting and classifying fish from sonar echo returns. Preliminary results are encouraging; a simple neural network was able to identify up to 86% of the test samples. When the identification problem was divided into three subproblems, over 93% of the samples were identified correctly. This success rate was found to be superior to both discriminant analysis and nearest neighbor techniques. Future research activities are discussed.<>
{"title":"The potential of a neural network based sonar system in classifying fish","authors":"P. Patrick, N. Ramani, W.G. Hanson, H. Anderson","doi":"10.1109/ICNN.1991.163352","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163352","url":null,"abstract":"The authors explore the potential of a neural network based system for detecting and classifying fish from sonar echo returns. Preliminary results are encouraging; a simple neural network was able to identify up to 86% of the test samples. When the identification problem was divided into three subproblems, over 93% of the samples were identified correctly. This success rate was found to be superior to both discriminant analysis and nearest neighbor techniques. Future research activities are discussed.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"84 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115381905","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 : 1991-08-15DOI: 10.1109/ICNN.1991.163342
A. Westerman
A lightweight, direct-drive undersea testbed manipulator arm was configured for integration and subsequent evaluation of neural network technologies. The author reports them initial results of an artificial neural network model used to control this undersea manipulator. An iterative trajectory generator for the manipulator (constrained to planar motion) using a backpropagation network is described. It provided the intermittent desired joint angles given the relative position information about the arm and the target. This work built upon the extended work of D. Sobajic and L. Pao, (1988). The author discusses a preliminary neural network architecture which learns the internal and controller model for the undersea manipulator arm. This control structure was inspired by the work of D. Nguyen and B. Widrow, (1990). Although this work is still underway, preliminary tests are encouraging, and are aimed at satisfying the adaptive capability necessary for operating in an unstructured ocean environment.<>
{"title":"Neural network control of a robotic manipulator arm for undersea applications","authors":"A. Westerman","doi":"10.1109/ICNN.1991.163342","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163342","url":null,"abstract":"A lightweight, direct-drive undersea testbed manipulator arm was configured for integration and subsequent evaluation of neural network technologies. The author reports them initial results of an artificial neural network model used to control this undersea manipulator. An iterative trajectory generator for the manipulator (constrained to planar motion) using a backpropagation network is described. It provided the intermittent desired joint angles given the relative position information about the arm and the target. This work built upon the extended work of D. Sobajic and L. Pao, (1988). The author discusses a preliminary neural network architecture which learns the internal and controller model for the undersea manipulator arm. This control structure was inspired by the work of D. Nguyen and B. Widrow, (1990). Although this work is still underway, preliminary tests are encouraging, and are aimed at satisfying the adaptive capability necessary for operating in an unstructured ocean environment.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127168571","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 : 1991-08-15DOI: 10.1109/ICNN.1991.163370
G. Carpenter, S. Grossberg, J. Reynolds
Summary form only given. The authors introduced a neural network architecture, called ARTMAP, that autonomously learns to classify arbitrarily many, arbitrarily ordered vectors into recognition categories based on predictive success. This supervised learning system is built up from a pair of adaptive resonance theory modules (ART/sub a/ and ART/sub b/) that are capable of self-organizing stable recognition categories in response to arbitrary sequences of input patterns. Tested on a benchmark machine learning database in both online and offline simulations, the ARTMAP system learns orders of magnitude more quickly, efficiently, and accurately than alternative algorithms, and achieves 100% accuracy after training on less than half of the input patterns in the database.<>
{"title":"ARTMAP: supervised real-time learning and classification of nonstationary data by a self-organizing neural network","authors":"G. Carpenter, S. Grossberg, J. Reynolds","doi":"10.1109/ICNN.1991.163370","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163370","url":null,"abstract":"Summary form only given. The authors introduced a neural network architecture, called ARTMAP, that autonomously learns to classify arbitrarily many, arbitrarily ordered vectors into recognition categories based on predictive success. This supervised learning system is built up from a pair of adaptive resonance theory modules (ART/sub a/ and ART/sub b/) that are capable of self-organizing stable recognition categories in response to arbitrary sequences of input patterns. Tested on a benchmark machine learning database in both online and offline simulations, the ARTMAP system learns orders of magnitude more quickly, efficiently, and accurately than alternative algorithms, and achieves 100% accuracy after training on less than half of the input patterns in the database.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126517149","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 : 1991-08-15DOI: 10.1109/ICNN.1991.163337
Z. Lin, S. Chittajallu, S. Kayalar, D. Wong, H. Yurtseven
Constant delay-sensitive neurons (CDNs) and tracking neurons (TNs) function as delay-dependent multipliers for cross-correlation processing of biosonar signals in the auditory cortex of the FM bat, Myotis lucifugus. Models of these two kinds of neurons using artificial neural networks (ANNs) which implement the back-propagation algorithm are presented. The ANNs were trained using data collected from neurophysiological experiments with an awake bat. Nonlinear transformations and parameters used in the models are discussed. An ANN model is presented for CDNs and another for TNs. The dynamic responses obtained from these models are observed to be comparable with the recorded signals of FM bats during actual hunting.<>
{"title":"Modeling constant best delay-sensitive neurons and tracking neurons in the auditory cortex of the FM bat with a back-propagation neural network","authors":"Z. Lin, S. Chittajallu, S. Kayalar, D. Wong, H. Yurtseven","doi":"10.1109/ICNN.1991.163337","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163337","url":null,"abstract":"Constant delay-sensitive neurons (CDNs) and tracking neurons (TNs) function as delay-dependent multipliers for cross-correlation processing of biosonar signals in the auditory cortex of the FM bat, Myotis lucifugus. Models of these two kinds of neurons using artificial neural networks (ANNs) which implement the back-propagation algorithm are presented. The ANNs were trained using data collected from neurophysiological experiments with an awake bat. Nonlinear transformations and parameters used in the models are discussed. An ANN model is presented for CDNs and another for TNs. The dynamic responses obtained from these models are observed to be comparable with the recorded signals of FM bats during actual hunting.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133113381","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 : 1991-08-15DOI: 10.1109/ICNN.1991.163331
A. Eapen
The author proposes the use of a neural network for detecting underwater targets in the presence of random noise. The neutral network is trained to analyze fixed time frames of the input signal to detect the presence or absence of the target, during which the network gets adapted to the local environment and learns to identify the features of the targets. A multilayer neural network is trained to correctly classify many example patterns with and without the target signal present. The back propagation learning rule is employed to update the weights on every presentation of input frames. Once the training is complete the network would be able to tell whether the input frame presented to it contains any target signature.<>
{"title":"Neural network for underwater target detection","authors":"A. Eapen","doi":"10.1109/ICNN.1991.163331","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163331","url":null,"abstract":"The author proposes the use of a neural network for detecting underwater targets in the presence of random noise. The neutral network is trained to analyze fixed time frames of the input signal to detect the presence or absence of the target, during which the network gets adapted to the local environment and learns to identify the features of the targets. A multilayer neural network is trained to correctly classify many example patterns with and without the target signal present. The back propagation learning rule is employed to update the weights on every presentation of input frames. Once the training is complete the network would be able to tell whether the input frame presented to it contains any target signature.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"81 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116473190","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 : 1991-08-15DOI: 10.1109/ICNN.1991.163357
L. Perlovsky
A maximum likelihood artificial neural system (MLANS) neural network is proposed for transient signal recognition. The MLANS learning efficiency greatly exceeds that of other neural networks and is approaching the information-theoretical limit on performance of any neural network or algorithm. The MLANS operates on a two-dimensional representation of the signal in either the short-term spectral or the Wigner transform domain. The first layer of the network uses structured second-order neurons to estimate the signal model from training data. A second layer performs optimal multimodal Bayes classification. Learning efficiency approaching the information-theoretical limit is achieved in each layer of the MLANS.<>
{"title":"Model based classification of transient signals using the MLANS neural network","authors":"L. Perlovsky","doi":"10.1109/ICNN.1991.163357","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163357","url":null,"abstract":"A maximum likelihood artificial neural system (MLANS) neural network is proposed for transient signal recognition. The MLANS learning efficiency greatly exceeds that of other neural networks and is approaching the information-theoretical limit on performance of any neural network or algorithm. The MLANS operates on a two-dimensional representation of the signal in either the short-term spectral or the Wigner transform domain. The first layer of the network uses structured second-order neurons to estimate the signal model from training data. A second layer performs optimal multimodal Bayes classification. Learning efficiency approaching the information-theoretical limit is achieved in each layer of the MLANS.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"140 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1991-08-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124346781","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：