By Topic

Neural Networks, 1990., 1990 IJCNN International Joint Conference on

Date 17-21 June 1990

Filter Results

Displaying Results 1 - 25 of 430
  • IJCNN International Joint Conference on Neural Networks (Cat. No.90CH2879-5)

    Publication Year: 1990
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (231 KB)  

    The following topics are dealt with: applications; invertebrate neural networks; image processing; supervised learning; associative memory; unsupervised learning; sensation and perception; electrical neurocomputers; sensormotor control systems; optical neurocomputers; machine vision; robotics and control; neurodynamics; neurocognition; and optimization View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Stock market prediction system with modular neural networks

    Publication Year: 1990 , Page(s): 1 - 6 vol.1
    Cited by:  Papers (37)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (447 KB)  

    A discussion is presented of a buying- and selling-time prediction system for stocks on the Tokyo Stock Exchange and the analysis of internal representation. The system is based on modular neural networks. The authors developed a number of learning algorithms and prediction methods for the TOPIX (Tokyo Stock Exchange Prices Indexes) prediction system. The prediction system achieved accurate predictions, and the simulation on stocks trading showed an excellent profit View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Computation of proximity effect corrections in electron beam lithography by a neural network

    Publication Year: 1990 , Page(s): 7 - 14 vol.1
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (530 KB)  

    The proximity effect, caused by electron-beam backscattering during resist exposure, can be compensated for by appropriate local changes in the incident beam dose, but the optimal correction, found iteratively, requires a prohibitively long time for realistic pattern sizes. A neural network has been used to perform these corrections, resulting in a significant decrease in computation time. The correction was first computed for a small test pattern using an iterative method. This solution was used as a training set for an adaptive, feedforward neural network, using back-propagation learning. After training, the network computed the same correction as the iterative method, but in a much shorter time View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Fingerprint processing using back propagation neural networks

    Publication Year: 1990 , Page(s): 15 - 20 vol.1
    Cited by:  Papers (2)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1085 KB)  

    A system for the extraction of minutiae from gray-scale fingerprint images using back-propagation networks and Gabor filters is described. Good detection ratios and low false alarm rates were achieved. The importance of having good input representations of the image data is illustrated through variations in the performance of different trained networks. The usefulness of Gabor filters in the representation processing layer is demonstrated View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Connectivity strategies for higher-order neural networks applied to pattern recognition

    Publication Year: 1990 , Page(s): 21 - 26 vol.1
    Cited by:  Papers (12)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (426 KB)  

    Different strategies for non-fully connected HONNs (higher-order neural networks) are discussed, showing that by using such strategies an input field of 128×128 pixels can be attained while still achieving in-plane rotation and translation-invariant recognition. These techniques allow HONNs to be used with the larger input scenes required for practical pattern-recognition applications. The number of interconnections that must be stored has been reduced by a factor of approximately 200000 in a T/C case and ~2000 in a Space Shuttle/F-18 case by using regional connectivity. Third-order networks have been simulated using several connection strategies View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Applying a Hopfield-style network to degraded text recognition

    Publication Year: 1990 , Page(s): 27 - 32 vol.1
    Cited by:  Papers (5)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (486 KB)  

    A Hopfield-style network model was developed and analyzed in detail by the author. This model has the advantage of higher storage capacity and less interference between stored memories than the classical discrete Hopfield network. The model is applied to machine printed word recognition. Words to be recognized are stored as content-addressable memories. Word images are first processed by an OCR. The network is then used to postprocess the OCR decisions View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Automotive diagnostics using trainable classifiers: statistical testing and paradigm selection

    Publication Year: 1990 , Page(s): 33 - 38 vol.1
    Cited by:  Papers (3)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (462 KB)  

    An analysis is presented of the requirements for developing a practical trainable classifier to detect and identify faults in vehicle powertrain systems. An examination is made of requirements on the data sets used for training and testing and the criteria needed to select the most appropriate classifier for a particular family of problems. Empirical results supporting the authors' hypothesis are presented based on an analysis of two data sets drawn under rather different circumstances from test vehicles with faults introduced. Several different classifier forms are applied to these data sets, and their performance is evaluated. Despite similar performance on simple statistical tests, the classifies exhibit significant performance variations on more rigorous tests, and therefore viable criteria for selecting the most appropriate classifiers can be established View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Frequency selective surface design based on iterative inversion of neural networks

    Publication Year: 1990 , Page(s): 39 - 44 vol.1
    Cited by:  Papers (7)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (380 KB)  

    A novel approach is presented to solve a constrained inverse problem encountered in the design of frequency selective surfaces (FSSs). Due to the many-to-one nonlinear functional relationship between an FSS and its frequency response, there is no closed-form solution directly from the given desired frequency response to the corresponding surface. Therefore, to design an FSS for a given response, one has to search in the knowledge base through a laborious and tedious trial-and-error procedure. The authors' approach adopts an iterative regularized inversion technique, which starts with an inversion algorithm for multilayer perceptrons to generate the corresponding 2-D surface for the given desired frequency response. A constraint-satisfaction mechanism is then used to reshape the 2-D surface to satisfy the constraints, and the resulting surface is used as the initial point for the next inversion algorithm. This procedure is mathematically similar to the projection-onto-convex-set algorithm for constrained optimization problems View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Radial basis function classification of impulse radar waveforms

    Publication Year: 1990 , Page(s): 45 - 50 vol.1
    Cited by:  Papers (6)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (438 KB)  

    The radial basis function algorithm is used to classify impulse radar waveforms from asphalt-covered bridge decks. A brief description of the impulse radar scenario is given, and the radial basis function algorithm is summarized. The classification results obtained using the algorithm are presented. Excellent success was obtained, with classification accuracies up to 99%. Training the radial basis function classifier is faster and less complex than training a back-propagation-based network, since the simple least-mean-squares (LMS) algorithm can be used to obtain estimates of the optimum weight values View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Hebbian feature discovery improves classifier efficiency

    Publication Year: 1990 , Page(s): 51 - 56 vol.1
    Cited by:  Papers (7)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (381 KB)  

    Two neural network implementations of principal component analysis (PCA) are used to reduce the dimension of speech signals. The compressed signals are then used to train a feedforward classification network for vowel recognition. A comparison is made of classification performance, network size, and training time for networks trained with both compressed and uncompressed data. Results show that a significant reduction in training time, fivefold in the present case, can be achieved without a sacrifice in classifier accuracy. This reduction includes the time required to train the compression network. Thus, dimension reduction, as performed by unsupervised neural networks, is a viable tool for enhancing the efficiency of neural classifiers View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Speaker-independent phoneme recognition on TIMIT database using integrated time-delay neural networks (TDNNs)

    Publication Year: 1990 , Page(s): 57 - 62 vol.1
    Cited by:  Papers (7)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (479 KB)  

    A structure of neural networks (NNs) is described for speaker-independent and context-independent phoneme recognition. This structure is based on the integration of time-delay neural networks (TDNN) which have several TDNNs separated according to the duration of phonemes. As a result, the proposed structure deals with phonemes of varying duration more effectively. In the experimental evaluation of the proposed structure, 16 English vowel recognition was performed using 5268 vowel tokens picked from 480 sentences spoken by 140 speakers (98 males and 42 females) on the TIMIT (TI-MIT) database. The number of training tokens and testing tokens was 4326 from 100 speakers (69 males and 31 females) and 942 from 40 speakers (29 males and 11 females), respectively. The result was a 60.5% recognition rate (around 70% for a collapsed 13-vowel case), which was improved from 56% in the single TDNN structure, showing the effectiveness of the proposed structure's use of temporal information View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • INSIDE: a neuronet based hardware fault diagnostic system

    Publication Year: 1990 , Page(s): 63 - 68 vol.1
    Cited by:  Papers (6)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (407 KB)  

    An inertial navigation system interactive diagnostic expert (INSIDE) was developed for troubleshooting an avionic line-replaceable unit, the inertial navigation system. INSIDE was designed based on a neural network model called neural-logic network. The knowledge base can be constructed using a neural-logic network by learning from past cases recorded in the workshop log book. To complement the connectionist knowledge base, a flowchart module which captures the knowledge of troubleshooting flowcharts was also implemented as part of the system. During operation, if the connectionist module fails to derive the solution, the user will be directed to the flowchart module for guidance. After the case is solved, it can be captured as a new example to be acquired by the connectionist module. Besides providing an economical way for developing fault diagnostic systems in general, the learning process of the system highly resembles the way an expert acquires knowledge through experience View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Classifying impulse radar waveforms using principle components analysis and neural networks

    Publication Year: 1990 , Page(s): 69 - 74 vol.1
    Cited by:  Papers (2)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (497 KB)  

    A multilayer neural network, trained with the back-propagation algorithm, is used to classify impulse radar waveforms from asphalt-covered bridge decks. A strategy for determining the structure of a bridge deck by using principal components analysis to reduce the dimensionality of the input data is demonstrated, showing classification accuracies ranging between 95.6% and 100%. The results show that neural networks can be used to extract information about a bridge deck's structure when waveforms from the deck are presented to it. Once the network has been trained to recognize several different structures, it should be possible to obtain very accurate estimates about the specific deck structure. The neural network will eliminate the need for taking core samples from the bridge deck, and a truly nondestructive bridge-deck evaluation system will be realized View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Role of activation function on hidden units for sample recording in three-layer neural networks

    Publication Year: 1990 , Page(s): 69 - 74 vol.3
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (464 KB)  

    It is shown that the k hidden units with asymptotic activation function are able to transfer any given k+1 different inputs to linearly independent GHUVs (generated hidden unit vectors) by properly setting weights and thresholds. The number of hidden units with the LIT (linearly independent transformation) capability for the polynomial activation function is limited by the order of polynomials. For analytic asymptotic activation functions and given different inputs, the LIT is a generic capability and a probability 1 capability in setting weights and thresholds randomly. It is a generic and a probability 1 property for any random input if the weight and threshold setting has LIT capability for some k+1 inputs. For three-layer nets with k hidden units, in which the activation function is asymptotic and the output layer is without activation function, they are sufficient to record k+1 arbitrary real samples. It is probability 0 to record k+2 random real samples if the activation is a unit step function. This is true for the sigmoid function in the case of associative memory. These conclusions lead to a scheme for understanding associative memory in the three-layer networks View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Proposal for neural-network applications to fiber-optic transmission

    Publication Year: 1990 , Page(s): 75 - 80 vol.1
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (425 KB)  

    Artificial neural networks are applied to fiber-optic transmission. Two fiber-optic transmission techniques using neural networks are proposed. One is an optical WDM (wavelength division multiplexing) demultiplexer composed of a simple optical component and all electrical neural network. The other is a fiber-optic image-transmission technique using a multimode fiber and a neural network. In either technique, propagation modes, in an optical multimode guide, play an important role in the signal processing. Initial experimental results are presented for these techniques. The combination of optics and neural networks have, so far, produced only the concept of optical neural networks. The techniques described can be regarded as different approaches to this combination View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Incipient fault detection and diagnosis using artificial neural networks

    Publication Year: 1990 , Page(s): 81 - 86 vol.1
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (420 KB)  

    Fault is defined as degradation between 100% performance and complete failure. The authors demonstrate how an artificial neural network can detect and diagnose faults from online process data. A wide range of input patterns can be learned by artificial neural networks in the presence of noise by changing the interconnections of the nodes, their thresholds for activation, and their individual weights. Artificial neural networks are able to take inputs from the processes without knowing the process model, allowing representation of complex engineering systems which are difficult to model either with traditional model-based engineering or knowledge-based expert systems. A description is given of some of the characters of a neural network that are useful for fault discrimination in a chemical plant. It is shown that even when using noisy sensor data, the misclassification rate is nil View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A neural sorting network with O(1) time complexity

    Publication Year: 1990 , Page(s): 87 - 95 vol.1
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (486 KB)  

    A multilayer feedforward neural network is proposed to solve sorting problems. The network has O(n2) neurons and O(n2 ) links. The number of layers is fixed regardless of input size. Thus, the computation time of the network is independent of input size, and the sorting network has a time complexity of O(1) View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Rule extraction and validity domain on a multilayer neural network

    Publication Year: 1990 , Page(s): 97 - 100 vol.1
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (318 KB)  

    The authors discuss the extraction of logical rules from a multilayer neural network when building a validity domain for the network is possible. This approach appears to be efficient when the expertise domain is restricted and when explicit rules are not easy to formulate. The validity domain is described as a set of mathematical and logical constraints. The constraints specifying the validity domain are then included in the constraint-propagation methods that can be used for extracting equivalent logical rules from the neural network. A corpus of jurisprudence illustrates these conditions very effectively. The logical rules that are extracted can be used either to explain the functioning of the neural network or as a first step for designing an expert system View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Classifying seismic signals via RCE neural network

    Publication Year: 1990 , Page(s): 101 - 105 vol.1
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (338 KB)  

    A discussion is presented of the use of a restricted coulomb energy (RCE) neural network to perform target recognition based on seismic signals. The training and testing patterns are the seismic signals of helicopters, tracked vehicles, and wheeled vehicles. It is found that the medium- and high-frequency parts of the seismic signal carry critical information for target recognition. A contrast-enhancement technique is applied to the medium- and high-frequency parts of the power spectrum to improve the performance of the system View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A neural network trained to select aircraft maneuvers during air combat: a comparison of network and rule based performance

    Publication Year: 1990 , Page(s): 107 - 112 vol.1
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (474 KB)  

    Research to develop a neural network model that selects aircraft maneuvers in the domain of air-combat maneuvering is described. A methodology for converting rule-based systems into a neural network was established. A comparison between the neural network and a rule-based expert system was undertaken. Differences between the architectures were explored, and hypotheses as to causes of differential performance were made. Both models were compared with expert fighter pilots on a transfer task. The neural network agreed with maneuver selections made by expert fighter pilots 2.5 times more often than the rule-based system. These findings were explained in terms of the ability of neural nets to generalize maneuver selections to novel airspace conditions. Implications of these results were also discussed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Cooperation of neural nets for robust classification

    Publication Year: 1990 , Page(s): 113 - 120 vol.1
    Cited by:  Papers (8)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (599 KB)  

    The problem of building robust neural network classification algorithms that perform well on a wide variety of problems is addressed. Evidence of the fact that current procedures are only accurate on a restricted range of tasks is presented. The authors propose multimodule architectures based on the cooperation of two or more neural net techniques as a solution. This idea is illustrated by the cooperation of a multilayer perceptron and a learning vector quantization algorithm. It is shown that this method combines the advantages of its individual components and is much more robust. It is accurate for a large range of problems and is easy to tune. An algorithm that allows the direct training of this multimodule architecture is described. The use of this technique enhances performance when dealing with classification problems View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A neural network architecture for the decoding of long constraint length convolutional codes

    Publication Year: 1990 , Page(s): 121 - 126 vol.1
    Cited by:  Papers (4)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (367 KB)  

    A neural network decoder designed to provide a constant output of decoded data for long-constraint-length convolutional codes (K⩾11) is presented. With only local connections between neurons and digital EX-OR cells, direct hardware implementation in a VLSI ASIC (application-specific integrated circuit) is feasible. Decoder strategy is discussed along with toggling strategy. Architectural modifications to decode other code rates are also discussed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Neural computation for planning AND/OR precedence-constraint robot assembly sequences

    Publication Year: 1990 , Page(s): 127 - 142 vol.1
    Cited by:  Papers (3)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (863 KB)  

    The problem of finding AND/OR precedence-constraint assembly sequences for a set of n parts that construct a mechanical object using neural computation is discussed. The geometric constraints of the assembled object are transformed into the elements of the connection matrix which specifies the connection strength among neurons. A modified Hopfield network is used to tackle the AND/OR precedence-constraint assembly-sequence problem. The designed algorithm can accommodate various constraints and applications. Detailed algorithms and analysis, and examples and experiments are presented View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • NPS: a neural network programming system

    Publication Year: 1990 , Page(s): 143 - 148 vol.1
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (302 KB)  

    A neural network programming system called NPS is presented. It is portable and has the ability to deal with network problems in general. The main aim of the system is to support portability and model independence by facilitating the implementation of a range of neural network models on a range of hardware. NPS is based on a specialized neural network language called NIL, a machine-independent network specification language designed to map a spectrum of neural models onto a range of architectures View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A shunting inhibitory motion detector that can account for the functional characteristics of fly motion-sensitive interneurons

    Publication Year: 1990 , Page(s): 149 - 153 vol.1
    Cited by:  Papers (3)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (428 KB)  

    The multiplicative inhibitory motion detector (MIMD) has been proposed by the authors (Visual Communications and Image Processing IV, Proceedings of SPIE. 1199, 1229-1240, 1989). The applicability of this motion detector to the activity of motion-sensitive interneurons of the lobula plate, the posterior pat of the third visual ganglion in the fly's optic lobe, is investigated. In particular, it is demonstrated that an array of MIMDs can simulate the characteristics of transient and steady-state response and of contrast sensitivity functions for these neurons View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.