Open Access. Powered by Scholars. Published by Universities.®
Articles 31 - 43 of 43
Full-Text Articles in Engineering
Voice Recognition Using Neural Networks, Ganesh K. Venayagamoorthy, Viresh Moonasar, K. Sandrasegaran
Voice Recognition Using Neural Networks, Ganesh K. Venayagamoorthy, Viresh Moonasar, K. Sandrasegaran
Electrical and Computer Engineering Faculty Research & Creative Works
One solution to the crime and illegal immigration problem in South Africa may be the use of biometric techniques and technology. Biometrics are methods for recognizing a user based on unique physiological and/or behavioural characteristics of the user. This paper presents the results of ongoing work into using neural networks for voice recognition
Adaptive Critic Designs, Danil V. Prokhorov, Donald C. Wunsch
Adaptive Critic Designs, Danil V. Prokhorov, Donald C. Wunsch
Electrical and Computer Engineering Faculty Research & Creative Works
We discuss a variety of adaptive critic designs (ACDs) for neurocontrol. These are suitable for learning in noisy, nonlinear, and nonstationary environments. They have common roots as generalizations of dynamic programming for neural reinforcement learning approaches. Our discussion of these origins leads to an explanation of three design families: heuristic dynamic programming, dual heuristic programming, and globalized dual heuristic programming (GDHP). The main emphasis is on DHP and GDHP as advanced ACDs. We suggest two new modifications of the original GDHP design that are currently the only working implementations of GDHP. They promise to be useful for many engineering applications …
High Order Orthogonal Tensor Networks: Information Capacity And Reliability, Donald C. Wunsch, Y. M. Mirkes, Alexander N. Gorban
High Order Orthogonal Tensor Networks: Information Capacity And Reliability, Donald C. Wunsch, Y. M. Mirkes, Alexander N. Gorban
Electrical and Computer Engineering Faculty Research & Creative Works
Neural networks based on construction of orthogonal projectors in the tensor power of space of signals are described. A sharp estimate of their ultimate information capacity is obtained. The number of stored prototype patterns (prototypes) can many times exceed the number of neurons. A comparison with the error control codes is made
Backpropagation Of Accuracy, Donald C. Wunsch, M. Yu Senashova, Alexander N. Gorban
Backpropagation Of Accuracy, Donald C. Wunsch, M. Yu Senashova, Alexander N. Gorban
Electrical and Computer Engineering Faculty Research & Creative Works
We solve the problem: how to determine maximal allowable errors, possible for signals and parameters of each element of a network, proceeding from the condition that the vector of output signals of the network should be calculated with given accuracy? "Backpropagation of accuracy" is developed to solve this problem
Input Dimension Reduction In Neural Network Training-Case Study In Transient Stability Assessment Of Large Systems, S. Muknahallipatna, Badrul H. Chowdhury
Input Dimension Reduction In Neural Network Training-Case Study In Transient Stability Assessment Of Large Systems, S. Muknahallipatna, Badrul H. Chowdhury
Electrical and Computer Engineering Faculty Research & Creative Works
The problem in modeling large systems by artificial neural networks (ANN) is that the size of the input vector can become excessively large. This condition can potentially increase the likelihood of convergence problems for the training algorithm adopted. Besides, the memory requirement and the processing time also increase. This paper addresses the issue of ANN input dimension reduction. Two different methods are discussed and compared for efficiency and accuracy when applied to transient stability assessment.
Two Methods Of Neural Network Controlled Dynamic Channel Allocation For Mobile Radio Systems, Kelvin T. Erickson, Edward J. Wilmes
Two Methods Of Neural Network Controlled Dynamic Channel Allocation For Mobile Radio Systems, Kelvin T. Erickson, Edward J. Wilmes
Electrical and Computer Engineering Faculty Research & Creative Works
Two methods of dynamic channel allocation using neural networks are investigated. Both methods continuously optimize the mobile network based on changes in calling traffic. The first method uses backpropagation model predictions to aid the channel allocator. Each cell contains a backpropagation model which provides the channel allocator a call traffic prediction allowing the channel allocator to effectively optimize the network. The second method uses the same backpropagation models along with actor-critic models to perform the channel allocation. The actor-critics learn to model traffic activity between adjacent cells in real-time, and thereby learn to allocate channels dynamically between cells. The learning …
Conservative Thirty Calendar Day Stock Prediction Using A Probabilistic Neural Network, H. Tan, Donald C. Wunsch, Danil V. Prokhorov
Conservative Thirty Calendar Day Stock Prediction Using A Probabilistic Neural Network, H. Tan, Donald C. Wunsch, Danil V. Prokhorov
Electrical and Computer Engineering Faculty Research & Creative Works
We describe a system that predicts significant short-term price movement in a single stock utilizing conservative strategies. We use preprocessing techniques, then train a probabilistic neural network to predict only price gains large enough to create a significant profit opportunity. Our primary objective is to limit false predictions (known in the pattern recognition literature as false alarms). False alarms are more significant than missed opportunities, because false alarms acted upon lead to losses. We can achieve false alarm rates as low as 5.7% with the correct system design and parameterization.
Terrain Classification In Sar Images Using Principal Components Analysis And Neural Networks, Mahmood R. Azimi-Sadjadi, Saleem Ghaloum, R. Zoughi
Terrain Classification In Sar Images Using Principal Components Analysis And Neural Networks, Mahmood R. Azimi-Sadjadi, Saleem Ghaloum, R. Zoughi
Electrical and Computer Engineering Faculty Research & Creative Works
The development of a neural network-based classifier for classifying three distinct scenes (urban, park and water) from several polarized SAR images of San Francisco Bay area is discussed. The principal component (PC) scheme or Karhunen-Loeve (KL) transform is used to extract the salient features of the input data, and to reduce the dimensionality of the feature space prior to the application to the neural networks. Employing PC scheme along with polarized images used in this study, led to substantial improvements in the classification rates when compared with previous studies. When a combined polarization architecture is used the classification rate for …
A Hybrid Optoelectronic Art-1 Neural Processor, Donald C. Wunsch, T. P. Caudell
A Hybrid Optoelectronic Art-1 Neural Processor, Donald C. Wunsch, T. P. Caudell
Electrical and Computer Engineering Faculty Research & Creative Works
Summary form only given. A new implementation of ART-1 (adaptive resonance theory) has been proposed that efficiently combines optical and electronic devices. All parallel computations are performed by the optics, while serial operations are performed in electronics.
A Neural Architecture For Unsupervised Learning With Shift, Scale And Rotation Invariance, Efficient Software Simulation Heuristics, And Optoelectronic Implementation, Donald C. Wunsch, D. S. Newman, T. P. Caudell, R. A. Falk, C. David Capps
A Neural Architecture For Unsupervised Learning With Shift, Scale And Rotation Invariance, Efficient Software Simulation Heuristics, And Optoelectronic Implementation, Donald C. Wunsch, D. S. Newman, T. P. Caudell, R. A. Falk, C. David Capps
Electrical and Computer Engineering Faculty Research & Creative Works
A simple modification of the adaptive resonance theory (ART) neural network allows shift, scale and rotation invariant learning. The authors point out that this can be accomplished as a neural architecture by modifying the standard ART with hardwired interconnects that perform a Fourier-Mellin transform, and show how to modify the heuristics for efficient simulation of ART architectures to accomplish the additional innovation. Finally, they discuss the implementation of this in optoelectronic hardware, using a modification of the Van der Lugt optical correlator
An Optoelectronic Adaptive Resonance Unit, Donald C. Wunsch, T. P. Caudell, R. A. Falk, C. David Capps
An Optoelectronic Adaptive Resonance Unit, Donald C. Wunsch, T. P. Caudell, R. A. Falk, C. David Capps
Electrical and Computer Engineering Faculty Research & Creative Works
The authors demonstrate a hardware implementation of the adaptive resonance theory ART 1 neural network architecture. The optoelectronic ART1 unit, is a novel application of an old device. This device-the 4-f or Van der Lugt correlator-has historically been used as a fast pattern classifier. Usually the correlation operation is employed as a matched filter, so that a maximum correlation peak corresponds to a well-matched pattern. The device described also uses the large peaks, but takes specific advantage of the fact that a zero-shift correlation is mathematically equivalent to a two-dimensional inner product. The authors describe a promising method for emulating …
A Commodity Trading Model Based On A Neural Network-Expert System Hybrid, K. Bergerson, Donald C. Wunsch
A Commodity Trading Model Based On A Neural Network-Expert System Hybrid, K. Bergerson, Donald C. Wunsch
Electrical and Computer Engineering Faculty Research & Creative Works
Demonstrates a system that combines a neural network approach with an expert system to provide superior performance compared to either approach alone. Learning capability is provided in a software-based approach to commodity trading systems. The authors used the backpropagation network with some parameters selected experimentally. They used a human expert to implicitly define patterns, using hindsight, that an intelligent system might have been able to use for an accurate prediction. Desired outputs were found by a combination of observing the behavior of technical indices that normally precede a certain kind of market behavior, and by observing the actual market behavior …
An Industrial Application To Neural Networks To Reusable Design, Donald C. Wunsch, R. Escobedo, T. P. Caudell, S. D. G. Smith, G. C. Johnson
An Industrial Application To Neural Networks To Reusable Design, Donald C. Wunsch, R. Escobedo, T. P. Caudell, S. D. G. Smith, G. C. Johnson
Electrical and Computer Engineering Faculty Research & Creative Works
Summary form only given, as follows. The feasibility of training an adaptive resonance theory (ART-1) network to first cluster aircraft parts into families, and then to recall the most similar family when presented a new part has been demonstrated, ART-1 networks were used to adaptively group similar input vectors. The inputs to the network were generated directly from computer-aided designs of the parts and consist of binary vectors which represent bit maps of the features of the parts. This application, referred to as group technology, is of large practical value to industry, making it possible to avoid duplication of design …