Pub Date : 1991-08-15DOI: 10.1109/ICNN.1991.163367
T. Washburne, M. Okamura, D. Specht, W. A. Fisher
The probabilistic neural network processor (PNNP) is a custom neural network parallel processor optimized for the high-speed execution (three billion connections per second) of the probabilistic neural network (PNN) paradigm. The performance goals for the hardware processor were established to provide a three order of magnitude increase in processing speed over existing neural net accelerator cards (HNC, FORD, SAIC). The PNN algorithm compares an input vector with a training vector previously stored in local memory. Each training vector belongs to one of 256 categories indicated by a descriptor table, which is previously filled by the user. The result of the comparison/conversion is accumulated in bins according to the original training vector's descriptor byte. The result is a vector of 256 floating-point works that is used in the final probability density function calculations.<>
{"title":"The Lockheed probabilistic neural network processor","authors":"T. Washburne, M. Okamura, D. Specht, W. A. Fisher","doi":"10.1109/ICNN.1991.163367","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163367","url":null,"abstract":"The probabilistic neural network processor (PNNP) is a custom neural network parallel processor optimized for the high-speed execution (three billion connections per second) of the probabilistic neural network (PNN) paradigm. The performance goals for the hardware processor were established to provide a three order of magnitude increase in processing speed over existing neural net accelerator cards (HNC, FORD, SAIC). The PNN algorithm compares an input vector with a training vector previously stored in local memory. Each training vector belongs to one of 256 categories indicated by a descriptor table, which is previously filled by the user. The result of the comparison/conversion is accumulated in bins according to the original training vector's descriptor byte. The result is a vector of 256 floating-point works that is used in the final probability density function calculations.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"102 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":"117191735","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.1991.163332
B. Bourgeois, C. Walker
The authors investigate the use of neural networks for the direct estimation of image texture. Unlike previous approaches where networks are used to make decisions on feature vectors derived from traditional techniques, or where a network is trained to perform the function of a traditional technique, the proposed approach uses a network to directly model texture. The envisioned approaches to this method are described. Preliminary results of one-dimensional tests show that a neural network implementation is very adapt at recognizing irregular signals, even in the presence of added noise. This is intended to be applied in a Seafloor Acoustic Imagery via sidescan imagery.<>
{"title":"Texture estimation with neural networks","authors":"B. Bourgeois, C. Walker","doi":"10.1109/ICNN.1991.163332","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163332","url":null,"abstract":"The authors investigate the use of neural networks for the direct estimation of image texture. Unlike previous approaches where networks are used to make decisions on feature vectors derived from traditional techniques, or where a network is trained to perform the function of a traditional technique, the proposed approach uses a network to directly model texture. The envisioned approaches to this method are described. Preliminary results of one-dimensional tests show that a neural network implementation is very adapt at recognizing irregular signals, even in the presence of added noise. This is intended to be applied in a Seafloor Acoustic Imagery via sidescan imagery.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"25 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":"133434813","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.1991.163323
Y. Pao, T.L. Hemminger, D. J. Adams, S. Clary
Acoustic transients develop and fade away continually in ocean environments. Accordingly, detection and interpretation of these are complicated by the fact that detection and classification cannot be made on the basis of temporal snapshots alone. Interpretation of transients must rest on the processing and classification of entire episodes of such continuing signals. The authors describe experiments in the design and implementation of such an episodal associative classifier which makes concurrent use of neural network self-organization and supervised learning methodologies. This system has no difficulty classifying signals from within test data sets and is also fast, robust, adaptive, and well suited for a wide range of sequence lengths.<>
{"title":"An episodal neural-net computing approach to the detection and interpretation of underwater acoustic transients","authors":"Y. Pao, T.L. Hemminger, D. J. Adams, S. Clary","doi":"10.1109/ICNN.1991.163323","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163323","url":null,"abstract":"Acoustic transients develop and fade away continually in ocean environments. Accordingly, detection and interpretation of these are complicated by the fact that detection and classification cannot be made on the basis of temporal snapshots alone. Interpretation of transients must rest on the processing and classification of entire episodes of such continuing signals. The authors describe experiments in the design and implementation of such an episodal associative classifier which makes concurrent use of neural network self-organization and supervised learning methodologies. This system has no difficulty classifying signals from within test data sets and is also fast, robust, adaptive, and well suited for a wide range of sequence lengths.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"103 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":"132333917","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.1991.163340
I. Schiller, J. Draper
The authors discuss lessons learned on a neural autonomous simulator project that can be applied to autonomous underwater vehicles (AUVs). They developed a neural network (NN)-based unmanned air vehicle (UAV) navigation demonstration. The UAV simulation shows friendly flight corridors, enemy air-defense sites and the UAV mission targets. The UAV navigates in this hostile environment and reacts to unexpected threats. The study concentrated on the feasibility for noncomputer experts to prepare the UAVs for the specialized missions dictated by mission requirements and the battle situation, such as SAM sites and goal locations, corridors or way points. It was shown that NNs are successful in operating UAVs, and that the mission success rate is improved over fixed way point to way point flying. The simulation shows the potential for enhancing AUV survivability in hostile environments.<>
{"title":"Mission adaptable autonomous vehicles","authors":"I. Schiller, J. Draper","doi":"10.1109/ICNN.1991.163340","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163340","url":null,"abstract":"The authors discuss lessons learned on a neural autonomous simulator project that can be applied to autonomous underwater vehicles (AUVs). They developed a neural network (NN)-based unmanned air vehicle (UAV) navigation demonstration. The UAV simulation shows friendly flight corridors, enemy air-defense sites and the UAV mission targets. The UAV navigates in this hostile environment and reacts to unexpected threats. The study concentrated on the feasibility for noncomputer experts to prepare the UAVs for the specialized missions dictated by mission requirements and the battle situation, such as SAM sites and goal locations, corridors or way points. It was shown that NNs are successful in operating UAVs, and that the mission success rate is improved over fixed way point to way point flying. The simulation shows the potential for enhancing AUV survivability in hostile environments.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"142 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":"132374047","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 Standard Transient Data Set (STDS) Phase 1 data were used to design detection and classification algorithms. Two separate processing chains were constructed, using neural networks for the short-duration transients and conventional processing for tonals. The design activity emphasized the judicious matching of acoustic digital signal processing (DSP) and neural networks, plus the construction of optimized training sets. The resulting design achieved 92% correct classification of the events in the testing files (204 correct out of 221 total events), with only four false alarms in approximately 35 min of data.<>
{"title":"A neural network-based passive sonar detection and classification design with a low false alarm rate","authors":"F.L. Casselman, D.F. Freeman, D.A. Kerrigan, S.E. Lane, N. Millstrom, W.G. Nichols","doi":"10.1109/ICNN.1991.163326","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163326","url":null,"abstract":"The Standard Transient Data Set (STDS) Phase 1 data were used to design detection and classification algorithms. Two separate processing chains were constructed, using neural networks for the short-duration transients and conventional processing for tonals. The design activity emphasized the judicious matching of acoustic digital signal processing (DSP) and neural networks, plus the construction of optimized training sets. The resulting design achieved 92% correct classification of the events in the testing files (204 correct out of 221 total events), with only four false alarms in approximately 35 min of data.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"7 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":"125171515","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.1991.163336
C.H. Chen
Summary form only given as follows. Although there has been little progress in conventional statistical and syntactic pattern recognition, the reemerging activity in neural networks and in artificial intelligence (AI) knowledge-based systems has had great impact in sonar and other applications. This impact is noted in the context of active sonar signal classification.<>
{"title":"Recent advances in pattern recognition and their potential application in active sonar classification","authors":"C.H. Chen","doi":"10.1109/ICNN.1991.163336","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163336","url":null,"abstract":"Summary form only given as follows. Although there has been little progress in conventional statistical and syntactic pattern recognition, the reemerging activity in neural networks and in artificial intelligence (AI) knowledge-based systems has had great impact in sonar and other applications. This impact is noted in the context of active sonar signal classification.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"138 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":"128474547","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.1991.163350
R. Pap, C. Parten, M. Rich, M. Lothers, C. Thomas
The authors explored whether neural networks can improve telerobotic performance. The neural network design is based upon an innovative three-tier distributed control architecture. The neurocontroller was tested on two simulated two-joint robot arms which had different dynamics. The tests compared the performance of five controller configurations. The findings indicate that a decentralized adaptive neurocontroller performed as well as or better than standard adaptive and nonadaptive controllers. This approach to autonomous control and path planning circumvents the tradeoff between speed and desired levels of accuracy, stability, and robustness by generating optimal trajectories without sacrificing computational speed or robustness.<>
{"title":"Underwater robotic operations using a decentralized adaptive neurocontroller","authors":"R. Pap, C. Parten, M. Rich, M. Lothers, C. Thomas","doi":"10.1109/ICNN.1991.163350","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163350","url":null,"abstract":"The authors explored whether neural networks can improve telerobotic performance. The neural network design is based upon an innovative three-tier distributed control architecture. The neurocontroller was tested on two simulated two-joint robot arms which had different dynamics. The tests compared the performance of five controller configurations. The findings indicate that a decentralized adaptive neurocontroller performed as well as or better than standard adaptive and nonadaptive controllers. This approach to autonomous control and path planning circumvents the tradeoff between speed and desired levels of accuracy, stability, and robustness by generating optimal trajectories without sacrificing computational speed or robustness.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"432 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":"125760864","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.1991.163361
P. Zakarauskas, J.M. Ozard, P. Brouwer
Two feedforward neural networks with one hidden layer each were trained using a modified backpropagation algorithm to determine the position of an acoustic source in a waveguide. One network was trained to localize the source in depth while the other was trained independently to localize in range. The signal was preprocessed by decomposition along an orthogonal basis vector set in order to increase the robustness of the resulting trained network to uncertainties in the signal and environmental parameters. The output layer consisted of one unit for each possible range or depth of the source. The networks were trained with a signal-to-noise ratio (S/N) of 50 dB and tested with patterns generated with S/Ns of 50 dB and 0 dB. Unambiguous localization was achieved with the trained network at 50 dB S/N, but the localization was more sensitive to the added noise at 0 dB S/N than a perceptron trained with one output cell for each combination of range and depth.<>
使用改进的反向传播算法训练两个各有一个隐藏层的前馈神经网络，以确定声源在波导中的位置。一个网络被训练为深度定位源，而另一个网络被独立训练为范围定位。为了提高训练后的网络对信号和环境参数的不确定性的鲁棒性，对信号沿正交基向量集进行分解预处理。输出层由一个单元组成，每个单元代表源的可能范围或深度。网络以50 dB的信噪比(S/N)进行训练，并以50 dB和0 dB的信噪比生成的图案进行测试。在50 dB S/N下，训练的网络实现了明确的定位，但在0 dB S/N下，定位对附加噪声更敏感，而在距离和深度的每个组合上训练一个输出单元
{"title":"Artificial neural networks for simultaneous and independent range and depth discrimination in passive acoustic localization","authors":"P. Zakarauskas, J.M. Ozard, P. Brouwer","doi":"10.1109/ICNN.1991.163361","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163361","url":null,"abstract":"Two feedforward neural networks with one hidden layer each were trained using a modified backpropagation algorithm to determine the position of an acoustic source in a waveguide. One network was trained to localize the source in depth while the other was trained independently to localize in range. The signal was preprocessed by decomposition along an orthogonal basis vector set in order to increase the robustness of the resulting trained network to uncertainties in the signal and environmental parameters. The output layer consisted of one unit for each possible range or depth of the source. The networks were trained with a signal-to-noise ratio (S/N) of 50 dB and tested with patterns generated with S/Ns of 50 dB and 0 dB. Unambiguous localization was achieved with the trained network at 50 dB S/N, but the localization was more sensitive to the added noise at 0 dB S/N than a perceptron trained with one output cell for each combination of range and depth.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"20 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":"127699176","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.1991.163321
J. Solinsky, E.A. Nash
The authors focus on passive sonar applications which involve analyzing data with unknown signals. A general set of signal events (which are classified by a human aural analysis) are used for network training. The primary objective of the application is to discriminate between target and nontarget event categories. A ground truth (GT) and classical decision theory are used in assessing various neural-network (NN) classifiers operating on the DARPA Phase 1 data set. Changes in classifier operating point are shown to vary results between classifier type. These results show the importance of identifying the objective of the NN application before performance assessment is made.<>
{"title":"Neural-network performance assessment in sonar applications","authors":"J. Solinsky, E.A. Nash","doi":"10.1109/ICNN.1991.163321","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163321","url":null,"abstract":"The authors focus on passive sonar applications which involve analyzing data with unknown signals. A general set of signal events (which are classified by a human aural analysis) are used for network training. The primary objective of the application is to discriminate between target and nontarget event categories. A ground truth (GT) and classical decision theory are used in assessing various neural-network (NN) classifiers operating on the DARPA Phase 1 data set. Changes in classifier operating point are shown to vary results between classifier type. These results show the importance of identifying the objective of the NN application before performance assessment is made.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"44 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":"131717788","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.1991.163338
S. Silven
A neural network for performing data association in a multitarget tracking system is described. Computer simulations have been conducted, and the results are presented. The solution to the data association problem, and therefore the design of the neural network is based on the minimization of a properly defined energy function. The derivation of the energy function is presented. The scoring function to be optimized is the sum of the probabilities of measurement-to-track file associations. The latter are derivable from a Kalman filter, which maintains the track files. The simulations indicate the ability of the neural network to converge quickly to the optimal hypothesis, which has the maximum score, given a reasonable difference in score between the optimal and nearest suboptimal hypothesis.<>
{"title":"A neural network for data association in a multiple-target tracking system","authors":"S. Silven","doi":"10.1109/ICNN.1991.163338","DOIUrl":"https://doi.org/10.1109/ICNN.1991.163338","url":null,"abstract":"A neural network for performing data association in a multitarget tracking system is described. Computer simulations have been conducted, and the results are presented. The solution to the data association problem, and therefore the design of the neural network is based on the minimization of a properly defined energy function. The derivation of the energy function is presented. The scoring function to be optimized is the sum of the probabilities of measurement-to-track file associations. The latter are derivable from a Kalman filter, which maintains the track files. The simulations indicate the ability of the neural network to converge quickly to the optimal hypothesis, which has the maximum score, given a reasonable difference in score between the optimal and nearest suboptimal hypothesis.<<ETX>>","PeriodicalId":296300,"journal":{"name":"[1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering","volume":"20 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":"122209855","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：