检索结果-内蒙古大学图书馆

A NORM SELECTION CRITERION FOR THE GENERALIZED DELTA RULE

IEEE TRANSACTIONS ON NEURAL NETWORKS 1991年第1期2卷 125-130页

作者： BURRASCANO, P INFO-COM Department University of Roma La Sapienza Rome Italy

The derivation of a supervised training algorithm for a neural network implies the selection of a norm criterion which gives a suitable global measure of the particular distribution of errors. The present paper addresses this problem and proposes a correspondence between error distribution at the output of a layered feedforward neural network and L(p) norms. The generalized delta rule is investigated in order to verify how its structure can be modified in order to perform a minimization in the generic L(p) norm. Finally, the particular case of the Chebyshev norm is developed and tested.

关键词： Neural networks backpropagation algorithms Particle measurements Feedforward neural networks Multilayer perceptrons Dispersion Testing Least squares methods Multi-layer neural network Stochastic processes

来源：评论

学校读者我要写书评

暂无评论

A NEURAL-NETWORK PREESTIMATION FILTER FOR BAD-DATA DETECTION AND IDENTIFICATION IN POWER-SYSTEM STATE ESTIMATION

引用

ELECTRIC POWER SYSTEMS RESEARCH 1995年第2期34卷 127-134页

作者： SALEHFAR, H ZHAO, R Department of Electrical and Computer Engineering Clarkson University Potsdam NY 13699-5720 USA

Popular state estimation techniques in industry are mostly based on the weighted least squares (WLS) method and its derivatives. These estimators usually detect and identify multiple gross measurement errors by repeating a cycle of estimation-detection-elimination. It is rather time consuming for large systems. This paper presents a neural network preestimation filter to identify most forms of gross errors, including conforming bad data, in raw measurements before state estimation rather than afterwards. The proposed neural network model is trained to be a measurement estimator by using the correct measurements of typical system operating states. Once trained, the filter quickly identifies most forms of gross measurement errors simultaneously by comparing the square difference of the raw measurements and their corresponding estimated values with some given thresholds. System observability is maintained by replacing bad data with their reasonably accurate estimates. Using the proposed neural network preestimation filter, the efficiency of present state estimators is greatly improved. Results from several case studies are presented.

关键词： STATE ESTIMATION DATA PROCESSING NEURAL NETWORKS backpropagation algorithms

来源：评论

学校读者我要写书评

暂无评论

Supervised self-coding in multilayered feedforward networks

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS 1996年第5期7卷 1184-1195页

作者： Sarukkai, RR Computer Science Department University of Rochester Rochester NY USA

Supervised neural-network learning algorithms have proven very successful at solving a variety of learning problems. However, they suffer from a common problem of requiring explicit output labels. This requirement makes such algorithms implausible as biological models. In this paper, it is shown that pattern classification can be achieved, in a multilayered feedforward neural network, without requiring explicit output labels, by a process of supervised self-coding. The class projection is achieved by optimizing appropriate within-class uniformity, and between-class discernibility criteria, The mapping function and the class labels are developed together, iteratively using the derived self-coding backpropagation algorithm. The ability of the self-coding network to generalize on unseen data is also experimentally evaluated on real data sets, and compares favorably with the traditional labeled supervision with neural networks, However, interesting features emerge out of the proposed self-coding supervision, which are absent in conventional approaches. The further implications of supervised self-coding with neural networks are also discussed.

关键词： Intelligent networks Neural networks backpropagation algorithms Supervised learning Labeling Biological system modeling Pattern classification Multi-layer neural network Feedforward neural networks

来源：评论

学校读者我要写书评

暂无评论

Dynamic Block-Wise Local Learning Algorithm for Efficient Neural Network Training

引用

IEEE TRANSACTIONS ON VERY LARGE SCALE INTEGRATION (VLSI) SYSTEMS 2021年第9期29卷 1680-1684页

作者： Lee, Gwangho Lee, Sunwoo Jeon, Dongsuk Seoul Natl Univ Grad Sch Convergence Sci & Technol Seoul 08826 South Korea

In the backpropagation algorithm, the error calculated from the output of the neural network should backpropagate the layers to update the weights of each layer, making it difficult to parallelize the training process and requiring frequent off-chip memory access. Local learning algorithms locally generate error signals which are used for weight updates, removing the need for backpropagation of error signals. However, prior works rely on large, complex auxiliary networks for reliable training, which results in large computational overhead and undermines the advantages of local learning. In this work, we propose a local learning algorithm that significantly reduces computational complexity as well as improves training performance. Our algorithm combines multiple consecutive layers into a block and performs local learning on a block-by-block basis, while dynamically changing block boundaries during training. In experiments, our approach achieves 95.68% and 79.42% test accuracy on the CIFAR-10 and CIFAR-100 datasets, respectively, using a small fully connected layer as auxiliary networks, closely matching the performance of the backpropagation algorithm. Multiply-accumulate (MAC) operations and off-chip memory access also reduce by up to 15% and 81% compared to backpropagation.

关键词： Training Heuristic algorithms Neural networks backpropagation backpropagation algorithms Memory management Deep learning Artificial neural networks concurrent computing local learning supervised learning

来源：评论

学校读者我要写书评

暂无评论

TRAINING FULLY RECURRENT NEURAL NETWORKS WITH COMPLEX WEIGHTS

引用

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-ANALOG AND DIGITAL SIGNAL PROCESSING 1994年第3期41卷 235-238页

作者： KECHRIOTIS, G MANOLAKOS, ES Center for Communications and Digital Signal Processing (CDSP) Research and Graduate Studies Electrical and Computer Engineering Department Northeastern University Boston MA USA

In this brief paper, the Real Time Recurrent Learning (RTRL) algorithm for training fully recurrent neural networks in real time, is extended for the case of a recurrent neural network whose inputs, outputs, weights and activation functions are complex. A practical definition of the complex activation function is adopted and the complex form of the conventional RTRL algorithm is derived. The performance of the proposed algorithm is demonstrated with an application in complex communication channel equalization.

关键词： Recurrent neural networks Signal processing algorithms Communication channels Neural networks backpropagation algorithms State-space methods Limit-cycles Speech processing Adaptive equalizers Digital signal processing

来源：评论

学校读者我要写书评

暂无评论

IMPROVED RESULTS FOR COLOR ERROR REDUCTIONS USING NEURAL NETWORKS AND A FAST TRAINING ALGORITHM

引用

IEEE TRANSACTIONS ON CONSUMER ELECTRONICS 1993年第4期39卷 923-927页

作者： CRILLY, PB XU, SF Electrical and Computer Engineering Communications Information and Signal Processing Group University of Tennessee Knoxville Knoxville TN USA

In a previous paper, neural networks were able to correct color errors in television receivers. The results showed that 70% of these errors were reduced by various amounts. This paper will further investigate the use of neural networks for error reduction. A simpler network will be employed in conjunction with a recently developed fast training algorithm. The new results are significantly better than previously reported. It will be shown that virtually all of the color errors will be significantly reduced. This paper will also present the details of the training algorithm.

关键词： Neural networks Signal processing algorithms Color backpropagation algorithms Computer errors Neurons Error correction Computer networks TV receivers Manufacturing

来源：评论

学校读者我要写书评

暂无评论

A REAL-TIME LEARNING ALGORITHM FOR A MULTILAYERED NEURAL NETWORK BASED ON THE EXTENDED KALMAN FILTER

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 1992年第4期40卷 959-966页

作者： IIGUNI, Y SAKAI, H TOKUMARU, H KYOTO UNIV FAC ENGNDIV APPL SYST SCIKYOTO 606JAPAN

The extended Kalman filter (EKF) is well known as a state estimation method for a nonlinear system, and can be used as a parameter estimation method by augmenting the state with unknown parameters. A multilayered neural network is a nonlinear system having a layered structure, and its learning algorithm is regarded as parameter estimation for such a nonlinear system. In this paper, a new real-time learning algorithm for a multilayered neural network is derived from the EKF. Since this EKF-based learning algorithm approximately gives the minimum variance estimate of the linkweights, the convergence performance is improved in comparison with the backwards error propagation algorithm using the steepest descent techniques. Furthermore, tuning parameters which crucially govern the convergence properties are not included, which makes its application easier. Simulation results for the XOR and parity problems are provided.

关键词： Neural networks Multi-layer neural network Convergence Iterative algorithms Signal processing algorithms Nonlinear systems Parameter estimation Linear systems backpropagation algorithms Application software

来源：评论

学校读者我要写书评

暂无评论

INTELLIGENT PREDICTION AND CONTROL OF A LEADSCREW GRINDING PROCESS USING NEURAL NETWORKS

引用

COMPUTERS IN INDUSTRY 1993年第3期23卷 169-174页

作者： DING, H YANG, SZ ZHU, XB HUAZHONG UNIV SCI & TECHNOL DEPT MECH ENGN IWUHANHUBEI 430074PEOPLES R CHINA UNIV SASKATCHEWAN DEPT MECH ENGNSASKATOON S7N 0W0SASKATCHEWANCANADA

A new approach to intelligent prediction and control of manufacturing processes is studied by the use of neural networks in this paper. Based on a back-propagation mechanism a learning algorithm for multilayer neural networks, the ''one-by-one algorithm'', is presented which is particularly suitable for forecasting and process control. The differences between the one-by-one algorithm and the back-propagation algorithm in current use are clarified. Then an intelligent forecasting and control architecture for a leadscrew grinding process using the one-by-one algorithm is discussed.

关键词： MANUFACTURING PROCESS MODELING NEURAL NETWORKS backpropagation algorithms INTELLIGENT CONTROL LEADSCREW GRINDING PROCESS CONTROL

来源：评论

学校读者我要写书评

暂无评论

GRADIENT DESCENT LEARNING ALGORITHM OVERVIEW - A GENERAL DYNAMICAL-SYSTEMS PERSPECTIVE

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS 1995年第1期6卷 182-195页

作者： BALDI, P CALTECH DIV BIOLPASADENACA 91109

We give a unified treatment of gradient descent learning algorithms for neural networks using a general framework of dynamical systems, This general approach organizes and simplifies all the known algorithms and results which have been originally derived for different problems (fixed point/trajectory learning), for different model (discrete/continuous), for different architectures (forward/recurrent), and using different techniques (backpropagation, variational calculus, adjoint methods, etc.). The general approach can also be applied to derive new algorithms. We then briefly examine some of the complexity issues and limitations intrinsic to gradient descent learning. Throughout the paper, we focus on the problem of trajectory learning.

关键词： backpropagation algorithms Neurons Context modeling Biological neural networks Calculus Biological systems Organisms Joining processes Hebbian theory Propulsion

来源：评论

学校读者我要写书评

暂无评论

THEORETICAL INVESTIGATION OF THE ROBUSTNESS OF MULTILAYER PERCEPTRONS - ANALYSIS OF THE LINEAR CASE AND EXTENSION TO NONLINEAR NETWORKS

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS 1995年第3期6卷 560-571页

作者： KERLIRZIN, P REFREGIER, P THOMSON CSF CENT RECH LABOPT GRPF-91404 ORSAYFRANCE

In this paper, we address the problem of robustness in multilayer perceptrons. We present the main theoretical results in the case of linear neural networks with one hidden layer in order to go beyond the empirical study we previously made. We show that the robustness can greatly be improved and that even without decreasing performance in normal use. Finally, we show how this behavior, clearly demonstrated in the linear case, is an approximation of the behavior of nonlinear networks.

关键词： Robustness Multilayer perceptrons Computer aided software engineering Neural networks Degradation Image coding Nonhomogeneous media Performance loss Constraint optimization backpropagation algorithms

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：