# 1993 Third International Conference on Artificial Neural Networks

## Filter Results

Displaying Results 1 - 25 of 62
• ### Robot path planning using VLSI resistive grids

Publication Year: 1993, Page(s):163 - 167
Cited by:  Patents (1)
| | PDF (360 KB)

The authors (1991) have presented results from software simulations of path planning for mobile robots in 2-D environments using resistive grids. They showed that paths could be computed incrementally in real time with a serial computer (such as a DSP chip) for slowly-moving robots navigating in static environments. For rapidly-changing environments with dynamic obstacles, however, a hardware impl... View full abstract»

• ### Minimum variance control of a class of nonlinear plants with neural networks

Publication Year: 1993, Page(s):168 - 171
Cited by:  Papers (1)
| | PDF (332 KB)

In this paper the authors introduce a technique for nonlinear control based on minimum variance control ideas, originally introduced in Astrom (1970) for the linear case. They focus their attention on a class of discrete time models depending nonlinearly on the exogenous input. A minimum variance controller, made up of neural networks and linear blocks, is designed for these models. The quality of... View full abstract»

• ### Cascadability and in-situ learning for VLSI multi-layer networks

Publication Year: 1993, Page(s):56 - 60
| | PDF (472 KB)

Progress in analogue VLSI technology for neural networks has been steady rather than spectacular, but there now exists a number of well-proven designs which have been used to build small-scale demonstrators. There are at least two outstanding issues, however, which need to be addressed before real-world applications, such as intelligent sensors or autonomous robots, become possible: cascadability,... View full abstract»

• ### Using prototypes to solve problems in neural net classifiers

Publication Year: 1993, Page(s):195 - 199
| | PDF (360 KB)

It is shown that using the notion of prototype' in neural net classifiers can solve several problems that a nonlinear classifier might encounter. This has motivated the development of a new neural classifier named NNP. The author presents an empirical evaluation on the advantages of NNP over other classifiers View full abstract»

• ### Neural networks performing system identification for control applications

Publication Year: 1993, Page(s):172 - 176
Cited by:  Papers (1)  |  Patents (1)
| | PDF (356 KB)

In this paper the ability of a neural network to perform a multidimensional curve-fitting is used to let a multilayer perceptron system identify a nonlinear multivariable dynamic process. The identified model is the well-known innovation state space model (Kalmann predictor). The identification is based only on input/output measurements, so in fact the extended Kalmann filter problem is solved. Th... View full abstract»

• ### An adaptive Bayesian network for low-level image processing

Publication Year: 1993, Page(s):61 - 65
Cited by:  Papers (1)  |  Patents (4)
| | PDF (404 KB)

Probability calculus, based on the axioms of inference, is the only consistent scheme for performing inference; this is also known as Bayesian inference. The objects which this approach manipulates, namely probability density functions (PDFs), may be created in a variety of ways, but the focus of this paper is on the use of adaptive PDF networks. Adaptive mixture distribution (MD) networks are alr... View full abstract»

• ### Scene segmentation of natural images using texture features and back-propagation

Publication Year: 1993, Page(s):200 - 204
| | PDF (440 KB)

Knowledge of three-dimensional world is essential for many guidance and navigation applications. A sequence of images from an electro-optical sensor can be processed using optical flow algorithms to provide a sparse set of ranges as a function of azimuth and elevation. A natural way to enhance the range map is by interpolation. However, this should be undertaken with care since interpolation assum... View full abstract»

• ### Novel topographic' nonlinear feature extraction using radial basis functions for concentration coding in the artificial nose'

Publication Year: 1993, Page(s):95 - 99
Cited by:  Papers (1)
| | PDF (392 KB)

A radial basis function technique is discussed which generates a nonlinear feature extraction mapping in which an ordering (possibly subjective) or similarity of the original data is also maintained. The technique is motivated by a real-data example and discussed in the context of an artificial nose'-a chemical vapour analysis employing broad-band sensor arrays. The particular motivation is that ... View full abstract»

• ### Target recognition in infra-red imagery using neural networks and machine learning

Publication Year: 1993, Page(s):21 - 25
| | PDF (364 KB)

This paper describes work undertaken by British Aerospace (BAe) on the evaluation of neural network and machine learning classifier techniques for automatic recognition of land based targets in infra-red imagery. The input to the classifier was derived from a histogram segmentation process extracting regions of interest from infra-red (IR) imagery. A set of statistical features were calculated for... View full abstract»

• ### Comparing parameters selection methods and weights rounding techniques to optimize the learning in neural networks

Publication Year: 1993, Page(s):46 - 50
| | PDF (388 KB)

Neural network techniques can be used for the approximation of decision functions. In such a case, the function's parameters to be estimated are the synaptic weights of the network. A small amount of data available for training limits the number of parameters that can be correctly estimated. Furthermore, all weights are not necessarily significant. An interesting point is to be able to decide whic... View full abstract»

• ### Learning and prediction of nuclear radioactive properties with artificial neural networks

Publication Year: 1993, Page(s):186 - 190
| | PDF (324 KB)

In this paper artificial neural networks (ANNs) are trained with nuclear radioactive properties using backpropagation (BP) as the learning paradigm. These properties are the radioactive decay modes and decay-gamma energies. The trained networks are used for predicting these properties for nuclei not included in the training set. The results obtained by prediction on test sets are compared with the... View full abstract»

• ### A neural network motion predictor

Publication Year: 1993, Page(s):177 - 181
Cited by:  Patents (1)
| | PDF (244 KB)

In this paper, the authors present the results of applying an Elman recurrent net to the prediction of the motion of an object. This neural network predictor can be used in the robot navigation system developed by Meng and Picton (1992) as an online predictor. An Elman net, a fast learning algorithm and an online learning scheme using specialised learning are described. Finally, they present the s... View full abstract»

• ### Neural network based dynamic models

Publication Year: 1993, Page(s):257 - 261
Cited by:  Patents (1)
| | PDF (336 KB)

This paper discusses the use of neural networks for modelling linear or nonlinear dynamic systems using the input and output temporal signals of the system to be modelled. The characteristics of a particular model are discussed, and some improvements are suggested. Sztipanovits proposed the application of the backpropagation network and of a special linear filter component (1992), the resonator ba... View full abstract»

• ### Investigating the recognition of false patterns in backpropagation networks

Publication Year: 1993, Page(s):133 - 137
| | PDF (324 KB)

This paper shows how a simple independent set of units working in parallel with a conventional multilayer perceptron network architecture enhance the reliability of the network with respect to the rejection of misshapen patterns. It is shown that rejection of invalid characters can be significantly improved without any degradation of performance with respect to the recognition of valid patterns. O... View full abstract»

• ### Improving generalisation with Ockham's networks: minimum description length networks

Publication Year: 1993, Page(s):81 - 85
| | PDF (420 KB)

There exists a substantial problem in obtaining good generalisation performance in the application of artificial neural network technology where training data is limited. A number of current techniques aiming to improve generalisation are introduced from the perspective of the minimum description length (MDL) principle. These are quadratic weight decay, soft weight-sharing and the technique introd... View full abstract»

• ### Neural computing: a new route to software reliability

Publication Year: 1993, Page(s):66 - 70
| | PDF (496 KB)

Neural networks, parallel distributed processing, or connectionism has been the focus of a rising new paradigm in artificial intelligence with sometimes a claimed association with neuroanatomy. If a trained network is viewed simply as an implementation of a function, it has to be admitted that it is a nonconventional implementation and is obtained by a nonconventional route. Networks may then be v... View full abstract»

• ### On-line adaptation in robot control

Publication Year: 1993, Page(s):205 - 209
| | PDF (272 KB)

Neural nets offer the chance of controlling a system effectively without knowing the relations between the required and directly controlled properties of the system. The authors report the training and performance of neural nets in the control loops of a robot arm, which are able to adapt to changes in the robot dynamics and reoptimise the control functionality without requiring intervention from ... View full abstract»

• ### Parallel learning algorithms for principal component extraction

Publication Year: 1993, Page(s):267 - 271
| | PDF (396 KB)

In this paper, learning rules for a two-layered network consisting of N input units and M output units, with full connections between the two layers and full lateral connections between the output units, are proposed. The learning rules extract the principal components from a given input data set, i.e. the weight vectors of the network converge to the eigenvectors belonging to th... View full abstract»

• ### Valid generalization in radial basis function networks and modified Kanerva models

Publication Year: 1993, Page(s):100 - 104
| | PDF (348 KB)

The Vapnik-Chervonenkis (VC) dimension has in recent years been successfully applied to the analysis of generalization in artificial neural networks of various types. The author presents an investigation of the VC dimension of radial basis function networks and of a related quantity, called the growth function, of modified Kanerva models View full abstract»

• ### Modelling of a fermentation process using a time-delay neural network

Publication Year: 1993, Page(s):220 - 223
| | PDF (212 KB)

The use of time-delay neural networks for the modelling of a fermentation process is addressed in this paper. The networks are used for the estimation of state variables, such as the secondary product concentration and the residual nitrogen. The networks were trained with data generated from an industrial fermentation process. The networks were tested as one-step ahead predictors and as models. Th... View full abstract»

• ### Neural network processing of scattered light measurements in the detection of immiscible water pollutants

Publication Year: 1993, Page(s):282 - 285
| | PDF (240 KB)

Using optoelectronic instrumentation, a smart sensing system for the detection and characterisation of oil pollution in water is demonstrated. Near infrared optics are used to derive sensor signatures in response to changes in background type and pollution levels. The signatures consist of scattered light intensities measured at 5 different angles. The measurements are used as input training sets ... View full abstract»

• ### Nonlinear noise filtering with neural networks: comparison with Weiner optimal filtering

Publication Year: 1993, Page(s):143 - 147
Cited by:  Papers (6)
| | PDF (288 KB)

This paper reports the application of a multi layer perceptron for the filtering of noisy time-series signals. The signals employed for the investigation are frequency modulated sine waves corrupted by different kinds of Gaussian and non-Gaussian noise. In general, overlapping spectra of signal and noise require nontrivial solutions for removing the noise with minimal degradation of the signal. Fo... View full abstract»

• ### An original approach for the localization of objects in images

Publication Year: 1993, Page(s):26 - 30
Cited by:  Papers (1)  |  Patents (1)
| | PDF (412 KB)

Presents an algorithm for the detection of faces in images using shared-weight replicated neural networks. A neural net forms rough hypotheses about the position of faces. These hypotheses are then verified using a second neural network. The algorithm applies to images where the size of the faces is unknown a priori. The computational time which is necessary for the complete processing of an image... View full abstract»

• ### Realization of physiological eye movements by automatic selection of control laws using artificial neural network

Publication Year: 1993, Page(s):113 - 117
Cited by:  Patents (1)
| | PDF (348 KB)

Eye movements are realized in hardware by controlling an optic axis according to the movement types of an object and a head using an artificial neural network. Saccadic and smooth pursuit eye movements are realized as an autokinesis, and the compensatory eye movement as a reflex by a hardware mechanism. For basic optic axis movements, by which an object is caught in a central pit of the retina, an... View full abstract»

• ### Variable bit rate block truncation coding for image compression using Hopfield neural networks

Publication Year: 1993, Page(s):233 - 237
| | PDF (284 KB)

A Hopfield neural network based block truncation coding (BTC) technique is presented in this paper. For this scheme, BTC is formulated as the minimization of a cost function in which the bit map distributions for the blocks are explicitly included. It is explained that this cost function may also be interpreted as a measure of the block detail. Based on the observation of the final value of the co... View full abstract»