# 1989 First IEE International Conference on Artificial Neural Networks, (Conf. Publ. No. 313)

## Filter Results

Displaying Results 1 - 25 of 87
• ### On the significance of internal representations in neural networks

Publication Year: 1989
| | PDF (30 KB)

Summary form only given. For a straightforward materialization of internal representations, semantic networks are suggested. In their original form they comprise a graph structure with nodes and links. The nodes may stand for items or concepts (sets of attributes), whereas the links usually indicate relations. In view of the contemporary neurophysiological data, such a degree of specificity and sp... View full abstract»

• ### Estimating hidden units for two-layer perceptrons

Publication Year: 1989, Page(s):120 - 124
Cited by:  Papers (25)
| | PDF (241 KB)

A method of estimating the number of hidden units required by a two-layer perceptron learning binary mappings using back propagation of error signals is presented. In order to obtain an estimate of the number of hidden units for a fully connected net with n output units, it is necessary to obtain an estimate of the number of 'conflicts' contained in the individual binary responses that must be lea... View full abstract»

• ### Diffusion learning for the multilayer perceptron

Publication Year: 1989, Page(s):390 - 394
Cited by:  Papers (5)  |  Patents (1)
| | PDF (191 KB)

A method of nonconvex optimisation based on simulating diffusion processes is discussed and applied to learning in the multilayer perceptron. It is compared with simulated annealing and back propagation of error as an optimisation learning tool. Results indicate significant improvement on these traditional learning methods. The method may be easily implemented in existing hardware.<> View full abstract»

• ### Identifying and discriminating temporal events with connectionist language users

Publication Year: 1989, Page(s):284 - 286
Cited by:  Papers (3)
| | PDF (188 KB)

Connectionist learning algorithms related to back-propagation have proven so effective that recent work has seriously considered the possibility of developing systems which learn and use natural language rather than processing it. This approach is termed the study of connectionist language users'. The connectionist language user paradigm is applied to several studies of the perception, processing... View full abstract»

• ### Implementation of plasticity in MOS synapses

Publication Year: 1989, Page(s):33 - 36
Cited by:  Papers (1)
| | PDF (216 KB)

Discusses learning, multiplication and other algorithms for a fully-connected artificial neural network in which synaptic weights between neurons are represented by resistances, the output of a neuron producing an analogue voltage Vi (in the range O to +V0 volts) which is a function of a weighted sum of its inputs. The strength or weight Wij of the connections is ... View full abstract»

• ### Image compression with competing multilayer perceptrons

Publication Year: 1989, Page(s):404 - 405
Cited by:  Papers (4)
| | PDF (188 KB)

Simple three-layer perceptrons with linear units working in auto-association with a reduced number of hidden units are applied to the task of digitized image compression. First, an algorithm developed using several multilayer perceptrons in competition for the coding of a TV-image is explained. A theoretical interpretation in terms of principal component analysis is also developed. Then, a study o... View full abstract»

• ### On the training and the convergence of brain-state-in-a-box neural networks

Publication Year: 1989, Page(s):247 - 251
| | PDF (324 KB)

It is the aim of the paper to contribute to the understanding and applicability of brain-state-in-a-box neural networks. It is shown how asymmetric brain-state-in-a-box neural networks achieve a multiple objective optimization, generalizing the energy'-interpretation of symmetric neural networks. It is therefore expected that asymmetric neural networks will have interesting applications once the ... View full abstract»

• ### Extension of the Hamming neural network to a multilayer architecture for optical implementation

Publication Year: 1989, Page(s):280 - 283
| | PDF (188 KB)

Examines how the Hamming net can be extended to a three layer architecture by means of the matched filter formalism. When the convergence parameters are set below a defined upper limit the net always converges to the correct pattern and so offers a definite advantage over the Hopfield net. Moreover, the Hamming convergence parameters can be made to affect the threshold offset rather than its slope... View full abstract»

• ### A general purpose digital architecture for neural network simulations

Publication Year: 1989, Page(s):62 - 66
Cited by:  Papers (1)  |  Patents (16)
| | PDF (564 KB)

A general-purpose neuro-chip for the resolution and learning stages of neural computing is described. Parallelism on input neurons is achieved for output neuron states evaluation and for synaptic weights updating. The latter follows a Hebb-like local formula. The digital realization guarantees exact integer calculations. According to simulations and theoretical considerations, 16-bit precision on ... View full abstract»

• ### Self-organization based on the second maximum entropy principle

Publication Year: 1989, Page(s):12 - 16
Cited by:  Papers (3)
| | PDF (192 KB)

The article formulates an optimal mapping from a continuous onto a discrete random variable by introducing the second maximum entropy principle, complementary to the Gibbsian one. The mapping corresponds to the self-organization of a community of formal neurons. The derived properties of the interaction between neurons are similar to those in biological neural networks View full abstract»

• ### Optical flow estimation by using the artificial neural network under multi-layers

Publication Year: 1989, Page(s):76 - 80
Cited by:  Patents (1)
| | PDF (332 KB)

A Hopfield model for computing optical flow is presented. A set of features describing the local intensity structure along the principal directions is used to measure the matching between the two local neighborhoods in the successive frames. The energy function can be derived based on the match measure and regularized by adding the Tikhonov stabilizer of the smoothness constraints. This energy fun... View full abstract»

• ### Silicon implementations of neural networks

Publication Year: 1989, Page(s):27 - 32
Cited by:  Papers (1)
| | PDF (496 KB)

Synthetic neural networks can be implemented in silicon as computer simulations, as digital or analog integrated circuits, or in a hybrid analog/digital form. The largest computational load in a neural system is incurred by the weighted summation Tij where Vj is a neural state and Tij the matrix of synaptic weights. This paper reviews re... View full abstract»

• ### Two-level recognition of isolated word using neural nets

Publication Year: 1989, Page(s):90 - 94
Cited by:  Papers (1)
| | PDF (268 KB)

Describes a neural-net based isolated word recogniser that has a better performance on a standard multi-speaker database than the reference hidden Markov model recogniser. The complete neural net recogniser is formed from two parts: a front-end which transforms the complex acoustic specification of the speech into a simplified phonetic feature specification, and a whole-word discriminator net. Eac... View full abstract»

• ### Adaptive radial basis function nonlinearities, and the problem of generalisation

Publication Year: 1989, Page(s):171 - 175
Cited by:  Papers (34)  |  Patents (4)
| | PDF (348 KB)

The author and D.S. Broomhead developed (1988) the opinion that most current feedforward layered neural networks perform a curve fitting operation in a high-dimensional space. To create the analogy, it was necessary to generalise earlier papers' assumptions, and so a mechanism for choosing radial basis functions was needed. The method involves optimisation. It is concluded that nonlinear optimisat... View full abstract»

• ### Learning with inference cells

Publication Year: 1989, Page(s):399 - 403
| | PDF (192 KB)

This paper introduces a learning method for neural nets using logic nodes. This learning method is based on very simple rules which are local to each cell. It is based on an inference mechanism associated with each cell and a transmitting/receiving process of inter-cell message passing View full abstract»

• ### Hardware realisable models of neural processing

Publication Year: 1989, Page(s):242 - 246
Cited by:  Patents (7)
| | PDF (240 KB)

An identity recently proposed by Gorse and Taylor (1988) between a certain class of neural model introduced by Taylor (1972) and a piece of electronic hardware, the `probabilistic random access memory' (pRAM) holds out the possibility of mimicking physiological nets in hardware in just this way. The model has recently been extended by Gorse and Taylor to operate at time scales of the order of the ... View full abstract»

• ### Optimal visual tracking with artificial neural network

Publication Year: 1989, Page(s):275 - 279
Cited by:  Papers (1)
| | PDF (292 KB)

Neural network approach is applicable according to two reasons: (1) there is no need for exact theoretical solution; and (2) it comprises high degree of potential parallel processing. The paper describes visual tracking problem and topology of the proposed neural network, and outlines the results of the learning phase (off line processing). Visual tracking and classification (on line processing) a... View full abstract»

• ### The implementation of hardware neural net systems

Publication Year: 1989, Page(s):57 - 61
Cited by:  Papers (3)
| | PDF (268 KB)

Describes a multilayer pipelined digital architecture suitable for the implementation of large neural nets (LN) for vision applications. It can also be used to do some pre-filtering, such as pixel averaging, by setting weight values appropriately. A 1024 node processor with a clock rate of 10 MHz can operate on an input vector consisting of 32×32 8 bit pixels in 102 μs. It can therefore p... View full abstract»

• ### A comparative study of the Kohonen and multiedit neural net learning algorithms

Publication Year: 1989, Page(s):7 - 11
| | PDF (280 KB)

This paper presents a comparative evaluation of the multiedit/condensing and Kohonen neural net learning algorithms using a speaker-independent speech recognition problem as a test vehicle. Both approaches attempt to cover the subspaces associated with respective pattern classes by a small number of reference vectors for subsequent nearest neighbour classification of unknown patterns. Several impo... View full abstract»

• ### Neural networks for speech pattern classification

Publication Year: 1989, Page(s):292 - 296
Cited by:  Papers (1)
| | PDF (228 KB)

The method of radial basis functions, a neural network model based on an approach to multidimensional interpolation, is described. The authors report on the application of a radial basis functions network and a back-propagation network to a vowel labelling problem and compare the results with those obtained by a vector-quantised hidden Markov model and a multivariate Gaussian classifier View full abstract»

• ### Dynamic scheduling for feed-forward neural nets using transputers

Publication Year: 1989, Page(s):257 - 260
Cited by:  Papers (1)
| | PDF (200 KB)

The modeling of neural networks on conventional digital computers can be a very time consuming operation. The authors evaluate one way to ease this time problem by mapping the processes involved onto an array of parallel processors. The neural approach to computing is inherently parallel with a fine level of granularity. This is to some extent incompatible with commercially available parallel proc... View full abstract»

• ### A neural network implementation for real-time scene analysis

Publication Year: 1989, Page(s):71 - 75
Cited by:  Papers (1)
| | PDF (260 KB)

A prototype neural network of LSI electronic logic, coupled with a matrix of photodetectors is proposed to implement the preprocessing functions of image capture and region extraction. The structure is described, with its likely implementation strategy, and the full online scene analysis algorithm is explained, with some simulation results made on real and 3D computer generated images View full abstract»

• ### Linear interpolation with binary neurons

Publication Year: 1989, Page(s):23 - 26
| | PDF (208 KB)

A two-layer network of binary neurons is considered. After learning a finite number of input-output combinations, the network performs linear interpolation between these combinations at the macroscopic level of correlations. It is not necessary to separate learning phase and testing phase. The network can also be taught linear transformations. It is shown that by introducing a special interpretati... View full abstract»

• ### Comparison of neural and conventional classifiers on a speech recognition problem

Publication Year: 1989, Page(s):86 - 89
Cited by:  Papers (3)
| | PDF (248 KB)

The authors set out to compare several different neural network and other methods on a common dataset relevant to automatic speech recognition. The particular problem chosen (speaker-independent EE-set recognition) is rather specialised, but particularly difficult View full abstract»

• ### Bearing estimation using neural optimisation methods

Publication Year: 1989, Page(s):129 - 133
Cited by:  Papers (1)
| | PDF (292 KB)

The bearing estimation problem is concerned with determining the directions of sources radiating an array of sensors, in the presence of additive noise. The authors have mapped the bearing estimation problem onto the Lyapunov energy function of the Hopfield model neural network. However, Hopfield model implements a gradient descent algorithm, and in common with all such algorithms, it is liable to... View full abstract»