检索结果-内蒙古大学图书馆

Proceedings of the 3rd Workshop on neural networks: Academic/Industrial/NASA/Defense

作者： Tam, David C. Univ of North Texas Denton United States

ISBN: (纸本)1565550072

A generic neural simulator, called MacNeuron, is implemented using an object-oriented programming approach to simulate the electrophysiological properties of neurons and interconnected neuronal networks. The neuronal model is based on the compartmental model where electrical properties of a subpart of neuron are compartmentalized and then 'link' together to reconstitute the whole neuron. Similarly, the properties of a neural network are reconstructed by connecting the neurons together to form a network. The neural simulator is designed with a generalizable philosophy in mind that will enable users to make additions and modifications to the existing neurophysiological models. Object-oriented approach is used extensively to encapsulate the similarities and differences between different neurons and their components by various objects. The object-oriented paradigm is used not only in constructing the neuronal anatomical hierarchy but also in computing the mathematical equations governing the electrical properties of the neurons. Thus, a truly generic and generalizable model for simulating neuronal properties can be conceived.

关键词： neural networks

来源：评论

学校读者我要写书评

暂无评论

Hybrid systems approaches with ascertainment learning in neural networks

Hybrid systems approaches with ascertainment learning in neu...

引用

Proceedings of the 3rd Workshop on neural networks: Academic/Industrial/NASA/Defense

作者： Reilly, Kevin D. Hayashi, Yoichi Villa, Mark F. Univ of Alabama at Birmingham United States

ISBN: (纸本)1565550072

neural network (NN) system which achieve maturity with relatively little training make good candidates for hybridizing with other forms of computing. Several terms may be used to describe the kind of learning to which we refer: one-time;exposure learning, which occurs in several prominent models;imprinting;or, with discernment, ascertainment learning, in which a certain but by no means significant amount of experience is required to capture knowledge. One reason such systems are good candidates for hybrids is that needs to transfer information among portions of the system, e.g., from a NN to an expert system or vice versa, can be accepted even if they are frequent. Other reasons include such systems' relative ease in management so that maintaining several of them simultaneously is not burdensome. This is particularly nice if the systems are distributed over a network of processors. This paper extends approaches taken earlier with hybrids, two cases of which we review briefly: 1) employing an expert system in conjunction with a competitive NN model and illustrating a form of cooperative behavior between a neural network and a symbolic conventional system;2) giving a Hopfield Net an added (conventional) numerical co-processor (filter) which helps govern input into the net. These systems reside under an umbrella of 'barrel' systems, also reported on earlier. Principal illustrations involve: 1) a combined BAM and Hamming Net system, which illustrates some advantages gained in a dual coding scheme;2) a stochastic form of Hamming Net;3) a BAM System's matrix allocations subjected to filter actions operating on memory matrices using a combination of intuitive and 'numerical recipe' computations.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

Boundary detection by artificial neural network

Boundary detection by artificial neural network

引用

International Joint Conference on neural networks (IJCNN)

作者： Kwok-Wai Cheung Tong Lee R.T. Chin Department of Electronic Engineering Chinese University of Hong Kong Sha Tin Hong Kong China Department of Computer Science Hong Kong University of Science and Technology Hong Kong China

"Active contour model-Snake" firstly suggested by Kass et al. (1988), is a boundary detection scheme, which is known to be very effective for detecting boundary problematic to existing classical schemes. However, the requirement of heavy computation limited its practical application. neural network, having a massively parallel architecture and being capable of processing huge amount of information in parallel manner, provides an alternative platform for real-time processing. In this paper, the "Snake" formulation is first mapped to a generalized higher-order Hopfield network and finally a tunneling network, an alternative neural network suggested by Cheung and Lee (1992), is adopted for the "Snake" boundary detection scheme. Simulation performed manifests its feasibility and it's found that the solution obtained is better than some existing "Snake" implementation.

关键词： Artificial neural networks neural networks Deformable models Active contours Tunneling Computer vision optical computing Computer science Computer architecture Hopfield neural networks

来源：评论

学校读者我要写书评

暂无评论

A time-frequency segmental neural network for phoneme recognition

A time-frequency segmental neural network for phoneme recogn...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： A. Basu T. Svendsen Department of Telecommunications Norwegian Institute of Technology Trondheim Norway

The authors propose a time-frequency segmental neural network (TFSNN) which classifies phonemes according to the two-dimensional time frequency distribution of the whole phonetic segment. It uses a network architecture similar to those used for optical character recognition (OCR) to provide local shift invariance along both the time and the frequency axis. The TFSNN can be used in place of a segmental neural network (SNN) in a hybrid hidden Markov model (HMM) artificial neural network (ANN) system for automatic speech recognition as it shows significantly better performance than the SNN. The training times for the TFSNN is also smaller as it employs very few connection weights compared with the SNN.< >

关键词： Time frequency analysis neural networks Hidden Markov models Artificial neural networks Speech recognition optical character recognition software Feature extraction optical computing optical network units Automatic speech recognition

来源：评论

学校读者我要写书评

暂无评论

Investigation of contrast enhancement by numerical methods for an optical cellular neural network

Investigation of contrast enhancement by numerical methods f...

引用

Midwest Symposium on Circuits and Systems (MWSCAS)

作者： P.P. Bey D.J. Yonce R.W. Newcomb Center for Bio/Molecular Science and Engineering Naval Research Laboratory Inc. WA USA Microsystems Laboratory University of Maryland College Park MD USA

A numerical model of a nearest-neighbor interconnected cellular neural network is described. The cells consist of optically driven autonulling circuits that generate analog output signals. For the numerical analysis we demonstrate that the 2D array provides lateral inhibition properties. Control of the image enhancement is obtained through the the weights associated with a 2D convolution operator.< >

关键词： optical computing optical fiber networks Cellular neural networks Bridge circuits Integrated circuit interconnections optical feedback optical filters Biomedical optical imaging Numerical models Signal generators

来源：评论

学校读者我要写书评

暂无评论

Decentralized Hopfield neural network applied to maintenance scheduling of generating units in power systems

Decentralized Hopfield neural network applied to maintenance...

引用

1993 Third International Conference on Artificial neural networks

作者： K. Yoshimoto K. Yasuda R. Yokoyama B.J. Cory Tokyo Metropolitan University Japan

In this paper, an efficient and practical method for solving a maintenance scheduling problem based on a neuro-computing approach is proposed. The objective of this problem is to equalize the capacity reserve over planing periods. The dynamical canonical network for nonlinear programming, into which is incorporated an improved Hopfield neural network for linear programming by Kennedy and Chua (1988), is modified to apply the neural network to 0-1 integer programming. To apply this approach to practical systems, a saving of computation burden has been attempted by decomposing it into a number of subproblems (that is, the maintenance period over one year is decomposed into several sub-periods) and applying a neural network to the respective subproblems. By introducing the decomposition technique and partial modification procedure, efficient and flexible maintenance scheduling of generating units has been obtained. The proposed approach based on the extended Hopfield neural network was applied to two kinds of power system models; medium-scale (15 generators and 24 periods) and practical-scale (60 generators and 52 periods). From the results of simulation, the effectiveness of the present approach has been verified as a fast-approximation solution method for the maintenance scheduling in medium or large scale power systems.< >

关键词：

来源：评论

学校读者我要写书评

暂无评论

The PAPRICA SIMD array: Critical reviews and perspectives

The PAPRICA SIMD array: Critical reviews and perspectives

引用

International Conference on Application Specific Array Processors

作者： F. Gregoretti C. Sansoe L.M. Reyneri A. Broggi G. Conte Department di Elettronica Politecnico di Turino Italy Department di Ingegneria dellľInformazione Universitià di Pisa Italy Department di Ingegneria dellľInformazione Università di Parma Italy

The PAPRICA project started in 1988 as an experimental VLSI architecture devoted to the efficient computation of data with two-dimensional structure. The main goal of the project is to develop a subsystem that could operate as an attached processing unit to a standard workstation and in perspective as a specialized processing module in dedicated systems devoted to low level image analysis, cellular neural networks emulation, DRC algorithms. The architecture has been extensively used for basic low level image analysis tasks up to optical flow computation and feature tracking, showing encouraging performance even in the first prototype version. The authors discuss the actual implementation and present a critical analysis of the project, allowing to identify some crucial points of PAPRICA design (and of array processors in general) that must be carefully considered in the case of redesign.< >

关键词： Computer architecture Image analysis optical computing Very large scale integration Standards development Workstations Cellular neural networks Emulation optical network units Image motion analysis

来源：评论

学校读者我要写书评

暂无评论

Morphological neural networks and image algebra in artificial perception systems

Morphological neural networks and image algebra in artificia...

引用

Image Algebra and Morphological Image Processing III

作者： Suarez Araujo, Carmen P. Ritter, Gerhard X. Univ. of Las Palmas de Gran Canaria Las Palmas de G. C. Canary Islands Spain

ISBN: (纸本)0819409421

Image algebra as a mathematical structure provides a much broader framework of neural computing. The matrix product in the basic equations of the current linear-based neural networks are furnished by the generalized matrix product obtaining new computational models as morphological neural networks (MNN). In this paper we propose a theoretic approach on the invariant perception. We also show that image algebra can be used not only in the field of image processing but in other areas related to artificial perception systems. Our study is based on both a general theory of neural network and the invariant perception by the cortex theory. The neural structures that we propose uphold both the architecture and functionality of the cortex. We present a neural network model for computing auditory homothetic invariances in accordance with a general framework in image algebra. The neuronal synthesis of this model is obtain using MNN theory with the binary operations the maximum and the multiplication in the neural network formulation. We also propose a second model which is achieved introducing a simple logarithmic transformation in the current model. In addition we propose an alternative MNN for computing homothetic invariances which arise from how the problems are formulated in the systems of artificial vision. This last neural network is appropriate to compute visual invariances when we process patterns defined in two dimension spaces.

关键词： neural networks

来源：评论

学校读者我要写书评

暂无评论

Quantum holography and neurocomputer architectures

引用

Journal of Mathematical Imaging and Vision 1992年第4期2卷 279-326页

作者： Schempp, Walter Lehrstuhl fuer Mathematik I University of Siegen Siegen D-5900 Germany

It is generally accepted among neuroscientists that the sensory cortex of the brain is arranged in a layered structure. Based on a unified quantum holographic approach to artificial neural network models implemented with coherent, hybrid optoelectronic, or analog electronic neurocomputer architectures, the present paper establishes a novel identity for the matching polynomials of complete bichromatic graphs which implement the intrinsic connections between neurons of local networks located in neural layers. © 1992 Kluwer Academic Publishers.

关键词： analog stimulus-response association cooperative synchronization Heisenberg nilpotent Lie group holography optical computing photonics

来源：评论

学校读者我要写书评

暂无评论

NET PRUNING OF MULTILAYER PERCEPTRON USING SEQUENTIAL CLASSIFICATION TECHNIQUE

NET PRUNING OF MULTILAYER PERCEPTRON USING SEQUENTIAL CLASSI...

引用

1ST CONF ON SCIENCE OF ARTIFICIAL neural networkS

作者： HUANG, KY YEN, HT National Chiao Tung Univ. (Taiwan)

ISBN: (纸本)0819408751

With the capabilities of parallel computation, distributed processing, and fault tolerance neural networks are employed widely in a number of research fields. Among the models of neural networks the single-layer perceptron and the multi-layer perceptron are the most popular ones used in supervised learning problems. However, there exists the redundant nodes that are insignificant for classification no matter which one of the two networks is trained to be a classifier. Although a net of a larger size usually has a faster learning rate, it results in an increase of forward computation complexity in either pattern recognizing or system relearning. In this paper, a new sequential classification model based on neural network is proposed. The model which combines the advantages of neural networks with the properties of the sequential classification is shown to have an encouraging performance for net pruning and feature reduction. In the experiments, two-class and m-class (m > 2) problems are implemented to prove the practicability of the new technique with a balance between the accuracy of pattern classification and the size of networks. In the conclusion, an overall discussion of the proposed model and technical comparisons with previous related research issues on net pruning are given.

关键词： Image classification neural networks Artificial neural networks FDA class I medical device development computing systems Feature selection Machine learning Performance modeling Software development Tolerancing

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：