检索结果-内蒙古大学图书馆

2ND INTERNATIONAL CONF ON ARTIFICIAL NEURAL NETWORKS

作者： ABRAHAM, IA BEDWORTH, MD BOOTH, CJM BOUNDS, DG MAYHEW, JEW POTTS, MAS THACKER, NA Research Initiative in Pattern Recognition United Kingdom

ISBN: (纸本)0852965311

We have developed a distributed multi-module system for realtime tracking of objects, and online training of neural network classifiers. The aim of the project was to build a system able to locate objects and to follow them and learn to recognise them with minimal supervision. The problem domain chosen was the tracking and recognition of three species of tropical fish swimming in an aquarium.

关键词： Neural Networks

来源：评论

学校读者我要写书评

暂无评论

SLOW ANNEALING VERSUS MULTIPLE FAST ANNEALING RUNS - AN EMPIRICAL-INVESTIGATION

引用

PARALLEL COMPUTING 1990年第2-3期16卷 269-272页

作者： DODD, N Research Initiative in Pattern Recognition Royal Signals and Radar Establishment Malvern WR14 3PS UK

The effects of splitting a single annealing run into several parallel shorter annealing runs is detailed for one specific application of combinatorial optimisation by simulated annealing. This form of parallelisation of the algorithm is extremely easy to implement since it requires no communication between the processors. It is shown that, for a considerable range of parallelisation, the probability of finding the global minimum is largely unchanged.

关键词： SIMULATED ANNEALING OPTIMIZATION PARALLEL COMPUTING

来源：评论

学校读者我要写书评

暂无评论

GRAPH MATCHING BY STOCHASTIC OPTIMIZATION APPLIED TO THE IMPLEMENTATION OF MULTI LAYER PERCEPTRONS ON TRANSPUTER NETWORKS

引用

PARALLEL COMPUTING 1989年第2期10卷 135-142页

作者： DODD, N Research Initiative in Pattern Recognition Royal Signals and Radar Establishments Malvern United Kingdom WR14 3PS

Multi Layer Perceptrons (MLPs) are typified by neuronal processes, requiring small processing power, connected to many other neuronal processes. Transputers are powerful processors but with only four links to connect to other transputers or associated hardware. Implementation of MLPs on transputers requires the allocation of neuronal processes to transputers and connection of transputers that will minimise traffic of signal through the transputer network. Stochastic optimisation, and in particular simulated annealing, has been found to produce minimal cost solutions to this problem. Furthermore the algorithms are themselves implementable on transputer networks.

关键词： graph isomorphism Graph matching optimisation parallel processing simulated annealing stochastic optimisation transputer

来源：评论

学校读者我要写书评

暂无评论

MULTISPECTRAL TEXTURE SYNTHESIS USING FRACTAL CONCEPTS

引用

IEEE TRANSACTIONS ON pattern ANALYSIS AND MACHINE INTELLIGENCE 1987年第5期9卷 703-707页

作者： DODD, N Research Initiative in Pattern Recognition RSRE St. Andrews Road Malvern WR1 3PS England

Natural textures are synthesized across several spectral bands using no more than a single second-order descriptor for each band. The Fractal model for monochrome textures is used as a null-hypothesis. The eye"s sensitivity to Fractal dimension under specific conditions is determined. Multispectral camouflage is generated to imitate a range of natural textures.","doi":"10.1109/TPAMI.1987.4767967","publicationTitle":"IEEE Transactions on pattern Analysis and Machine Intelligence","startPage":"703","endPage":"707","rightsLink":"http://***/AppDispatchServlet?publisherName=ieee&publication=0162-8828&title=Multispectral+Texture+Synthesis+Using+Fractal+Concepts&isbn=&publicationDate=Sept.+1987&author=Nigel+Dodd&ContentID=10.1109/TPAMI.1987.4767967&orderBeanReset=true&startPage=703&endPage=707&volumeNum=PAMI-9&issueNum=5","displayPublicationTitle":"IEEE Transactions on pattern Analysis and Machine Intelligence","pdfPath":"/iel5/34/4767950/***","keywords":[{"type":"IEEE Keywords","kwd":["Fractals","Brightness","Statistics","Fourier transforms","Algorithm design and analysis","Surface texture","research initiatives","pattern recognition","Painting","Image generation"]},{"type":"Author Keywords ","kwd":["texture","Camouflage","eigencomponent","Fractals","multispectral"]}],"allowComments":false,"pubLink":"/xpl/***?punumber=34","issueLink":"/xpl/***?isnumber=4767950","standardTitle":"Multispectral Texture Synthesis Using Fractal Concepts

关键词： Fractals Brightness Statistics Fourier transforms Algorithm design and analysis Surface texture research initiatives pattern recognition Painting Image generation Author Keywords texture Camouflage eigencomponent Fractals multispectral

来源：评论

学校读者我要写书评

暂无评论

Realtime tracking and identification of multiple objects in a cluttered environment

Realtime tracking and identification of multiple objects in ...

引用

1991 Second International Conference on Artificial Neural Networks

作者： I.A. Abraham M.D. Bedworth C.J.M. Booth D.G. Bounds J.E.W. Mayhew M.A.S. Potts N.A. Thacker Research Initiative Pattern Recognition UK

The authors have developed a distributed multi-module system for realtime tracking of objects, and online training of neural network classifiers. The aim of the project was to build a system which is able to locate objects and to follow them and learn to recognise them with minimal supervision. The problem domain chosen was the tracking and recognition of three species of tropical fish swimming in an aquarium.< >

关键词： Neural networks pattern recognition Real time systems

来源：评论

学校读者我要写书评

暂无评论

Node splitting: a constructive algorithm for feed-forward neural networks 91

Node splitting: a constructive algorithm for feed-forward ne...

引用

Proceedings of the 5th International Conference on Neural Information Processing Systems

作者： Mike Wynne-Jones Research Initiative in Pattern Recognition Great Malvern UK

ISBN: (纸本)9781558602229

A constructive algorithm is proposed for feed-forward neural networks, which uses node-splitting in the hidden layers to build large networks from smaller ones. The small network forms an approximate model of a set of training data, and the split creates a larger more powerful network which is initialised with the approximate solution already found. The insufficiency of the smaller network in modelling the system which generated the data leads to oscillation in those hidden nodes whose weight vectors cover regions in the input space where more detail is required in the model. These nodes are identified and split in two using principal component analysis, allowing the new nodes to cover the two main modes of each oscillating vector. Nodes are selected for splitting using principal component analysis on the oscillating weight vectors, or by examining the Hessian matrix of second derivatives of the network error with respect to the weights. The second derivative method can also be applied to the input layer, where it provides a useful indication of the relative importances of parameters for the classification task. Node splitting in a standard Multi Layer Perceptron is equivalent to introducing a hinge in the decision boundary to allow more detail to be learned. Initial results were promising, but further evaluation indicates that the long range effects of decision boundaries cause the new nodes to slip back to the old node position, and nothing is gained. This problem does not occur in networks of localised receptive fields such as radial basis functions or gaussian mixtures, where the technique appears to work well.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Optimisation of network structure using genetic techniques

Optimisation of network structure using genetic techniques

引用

International Joint Conference on Neural Networks (IJCNN)

作者： N. Dodd Royal Signals and Radar Establishment Research Initiative in Pattern Recognition Malvern UK

Arguments contending that it is necessary to use structured neural networks for the solution of certain problem types are presented. Structure is imposed on connectivity, activation functions, and other parameters of the network to simultaneously optimize generalization ability and compactness of the network. An analogy is made between the development of biological nervous systems from their genetic coding and the generation of artificial neural networks from a parametric description. Experiments which use genetic techniques to optimize network structure for a specific class of problem are described. Results which demonstrate the effectiveness of genetic optimization of network specifications in comparison with other optimization techniques are given. The parallel asynchronous implementation of genetic algorithms on a Sun-3 network is briefly described

关键词：

来源：评论

学校读者我要写书评

暂无评论

Recurrent networks for learning stochastic sequences

Recurrent networks for learning stochastic sequences

引用

International Joint Conference on Neural Networks (IJCNN)

作者： N. McCulloch Research Initiative in Pattern Recognition Royal Signals and Radar Establishment Malvern Worcestershire UK

Some experiments exploring the ability of networks to learn the underlying statistics of artificially generated temporal data are described. In one experiment, data generated by two simple Markov models were fed into a multilayer perceptron. The desired output was an indication of whether a transition out of one of the models had been made. The network produced a close approximation to the probability that a transition had just been made. In another experiment, hidden Markov models were used to generate the data. This made the determination of whether a transition had occurred much more difficult, and the network produced a much poorer approximation to the correct probability

关键词：

来源：评论

学校读者我要写书评

暂无评论

A sideways look at task decomposition

A sideways look at task decomposition

引用

IEE Colloquium on Knowledge Based Environments for Industrial Applications Including Co-Operating Expert Systems in Control

作者： C.J.M. Booth J.E.W. Mayhew Research Initiative in Pattern Recognition Glasgow UK AI Vision Research Unit Sheffield University Sheffield UK

Given a complex dynamic problem, one way to decompose the problem into subtasks is in terms of the behaviours needed to solve the problem. Such a behavioural decomposition is very natural, allowing the solution to be programmed on a distributed network of processors running in parallel. The subtasks are largely-but not wholly-independent, which reduces the amount of communication among them. An architecture, based on the subsumption architecture of R.A. Brooks (1986), is proposed to allow the implementation of such a decomposition. This architecture and a mobile robot, known as the Sprite, are simulated running in a 2-dimensional world. A controller for the Sprite is also simulated; it consists of four layers, and it allows the Sprite to avoid obstacles, to find and follow walls, and to make a simple map of what it encounters. It is believed that this architecture is also suitable for implementing large vision systems. In order to test this conjecture the architecture is being transferred onto a transputer network.< >

关键词： Distributed computing Mobile robots Parallel processing Robot programming

来源：评论

学校读者我要写书评

暂无评论

Speech recognition: statistical and neural information processing approaches 88

Speech recognition: statistical and neural information proce...

引用

Proceedings of the 2nd International Conference on Neural Information Processing Systems

作者： John S. Bridle Speech Research Unit and National Electronics Research Initiative in Pattern Recognition Royal Signals and Radar Establishment Malvern UK

Automatic Speech recognition (ASR) is an artificial perception problem: the input is raw, continuous patterns (no symbols!) and the desired output, which may be words, phonemes, meaning or text, is symbolic. The most successful approach to automatic speech recognition is based on stochastic models. A stochastic model is a theoretical system whose internal state and output undergo a series of transformations governed by probabilistic laws [1]. In the application to speech recognition the unknown patterns of sound are treated as if they were outputs of a stochastic system [18,2]. Information about the classes of patterns is encoded as the structure of these "laws" and the probabilities that govern their operation. The most popular type of SM for ASR is also known as a "hidden Markov model."

关键词：

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：