检索结果-内蒙古大学图书馆

Tree-structured Bayesian network learning with application to scene classification

ELECTRONICS LETTERS 2011年第9期47卷 540-541页

作者： Wang, Z. F. Wang, Z. H. Xie, W. J. Beijing Jiaotong Univ Sch Comp & Informat Technol Beijing 100044 Peoples R China

A tree-structured Bayesian network is one of the best probabilistic models for scene classification. A simple and successful learning algorithm for a tree-structured Bayesian network classifier is presented without taking arc reversal into account. A systematic performance study on the Lazebnik 15 dataset shows the proposed method is more efficient than the traditional learning scheme.

关键词： Optical, image and video signal processing learning (artificial intelligence) File organisation scene classification belief networks Lazebnik dataset Knowledge engineering techniques tree data structures tree-structured Bayesian network classifier learning algorithm image classification Computer vision and image processing techniques probabilistic models

来源：评论

学校读者我要写书评

暂无评论

Efficient perceptron learning using constrained steepest descent

引用

NEURAL NETWORKS 2000年第3期13卷 351-364页

作者： Perantonis, SJ Virvilis, V Natl Ctr Sci Res Demokritos Inst Informat & Telecommun Athens 15310 Greece

An algorithm is proposed for training the single-layered perceptron. The algorithm follows successive steepest descent directions with respect to the perceptron cost function, taking care not to increase the number of misclassified patterns. The problem of finding these directions is stated as a quadratic programming task, to which a fast and effective solution is proposed. The resulting algorithm has no free parameters and therefore no heuristics are involved in its application. It is proved that the algorithm always converges in a finite number of steps. For linearly separable problems, it always finds a hyperplane that completely separates patterns belonging to different categories. Termination of the algorithm without separating all given patterns means that the presented set of patterns is indeed linearly inseparable. Thus the algorithm provides a natural criterion for linear separability. Compared to other state of the art algorithms, the proposed method exhibits substantially improved speed, as demonstrated in a number of demanding benchmark classification tasks. (C) 2000 Published by Elsevier Science Ltd. All rights reserved.

关键词： perceptron learning algorithm quadratic programming feasible directions

来源：评论

学校读者我要写书评

暂无评论

Iterative learning Control Based on Nesterov Accelerated Gradient Method

引用

IEEE ACCESS 2019年 7卷 115836-115842页

作者： Gu, Panpan Tian, Senping Chen, Yangquan South China Univ Technol Sch Automat Sci & Engn Guangzhou 510640 Guangdong Peoples R China Univ Calif Merced Sch Engn MESA Lab Merced CA 95343 USA

Based on Nesterov accelerated gradient method, the problem of iterative learning control for a class of linear discrete-time systems is considered in this paper. Firstly, the iterative learning control problem of linear discrete-time systems is transformed into an iterative least-squares problem. Then, the Nesterov accelerated gradient method is introduced into the iterative learning control framework. Note that the Nesterov accelerated gradient learning algorithm has the capability of fast convergence. It is shown that the algorithm presented in this paper can guarantee the output tracking error converges to zero with rate O (1/k), where k is the iteration counter. Moreover, the monotonic convergence of the Nesterov accelerated gradient learning algorithm is analyzed and discussed. Finally, the effectiveness of the proposed method is verified by two simulation examples.

关键词： Iterative learning control Nesterov accelerated gradient method monotonic convergence learning algorithm

来源：评论

学校读者我要写书评

暂无评论

learning BY FUZZIFIED NEURAL NETWORKS

引用

INTERNATIONAL JOURNAL OF APPROXIMATE REASONING 1995年第4期13卷 327-358页

作者： ISHIBUCHI, H MORIOKA, K TURKSEN, IB UNIV TORONTO TORONTOONCANADA

We derive a general learning algorithm for training a fuzzified feedforward neural network that has fuzzy inputs, fuzzy targets, and fuzzy connection weights. The derived algorithm is applicable to the learning of fuzzy connection weights with various shapes such as triangular and trapezoid. First we briefly describe how a feedforward neural network can be fuzzified. inputs, targets, and connection weights in the fuzzified neural network can be fuzzy numbers. Next we define a cost function that measures the difference between a fuzzy target vector and an actual fuzzy output vector Then we derive a learning algorithm from the cost function for adjusting fuzzy connection weights. Finally we show some results of computer simulations.

关键词： FUZZIFICATION OF NEURAL NETWORKS FUZZY INPUTS FUZZY TARGETS FUZZY CONNECTION WEIGHTS learning algorithm FEEDFORWARD NEURAL NETWORKS

来源：评论

学校读者我要写书评

暂无评论

learning approach to nonlinear fault diagnosis: Detectability analysis

引用

IEEE TRANSACTIONS ON AUTOMATIC CONTROL 2000年第4期45卷 806-812页

作者： Polycarpou, MM Trunov, AB Univ Cincinnati Dept Elect & Comp Engn & Comp Sci Cincinnati OH 45221 USA

The learning approach to fault diagnosis provides a methodology for designing monitoring architectures which can he used for detection, identification and accommodation of failures in dynamical systems. This paper considers the issues of detectability conditions and detection time in a nonlinear fault diagnosis scheme based on the learning approach. First, conditions are derived to characterize the range of detectable faults. Then, nonconservative upper bounds are computed for the detection time of incipient and abrupt faults. It is shown that the detection time bound decreases monotonically as the values of certain design parameters increase. The theoretical results are illustrated by a simulation example of a second-order system.

关键词： detection time fault diagnosis learning algorithm nonlinear estimator on-line approximator

来源：评论

学校读者我要写书评

暂无评论

Low-dimensional-Structure Self-learning and Thresholding: Regularization Beyond Compressed Sensing for MRI Reconstruction

引用

MAGNETIC RESONANCE IN MEDICINE 2011年第3期66卷 756-767页

作者： Akcakaya, Mehmet Basha, Tamer A. Goddu, Beth Goepfert, Lois A. Kissinger, Kraig V. Tarokh, Vahid Manning, Warren J. Nezafat, Reza Beth Israel Deaconess Med Ctr Dept Med Boston MA 02215 USA Harvard Univ Sch Med Boston MA USA Beth Israel Deaconess Med Ctr Dept Radiol Boston MA 02215 USA Harvard Univ Sch Engn & Appl Sci Cambridge MA 02138 USA

An improved image reconstruction method from undersampled k-space data, low-dimensional-structure self-learning and thresholding (LOST), which utilizes the structure from the underlying image is presented. A low-resolution image from the fully sampled k-space center is reconstructed to learn image patches of similar anatomical characteristics. These patches are arranged into 'similarity clusters,' which are subsequently processed for dealiasing and artifact removal, using underlying low-dimensional properties. The efficacy of the proposed method in scan time reduction was assessed in a pilot coronary MRI study. Initially, in a retrospective study on 10 healthy adult subjects, we evaluated retrospective undersampling and reconstruction using LOST, wavelet-based I-1-norm minimization, and total variation compressed sensing. Quantitative measures of vessel sharpness and mean square error, and qualitative image scores were used to compare reconstruction for rates of 2,3, and 4. Subsequently, in a prospective study, coronary MRI data were acquired using these rates, and LOST-reconstructed images were compared with an accelerated data acquisition using uniform undersampling and sensitivity encoding reconstruction. Subjective image quality and sharpness data indicate that LOST outperforms the alternative techniques for all rates. The prospective LOST yields images with superior quality compared with sensitivity encoding or I-1-minimization compressed sensing. The proposed LOST technique greatly improves image reconstruction for accelerated coronary MRI acquisitions. Magn Reson Med 66:756-767,2011. (C)2011 Wiley-Liss, Inc.

关键词： compressed sensing accelerated imaging image reconstruction block matching learning algorithm cardiac MR coronary MRI

来源：评论

学校读者我要写书评

暂无评论

An adaptive penalty-based learning extension for the backpropagation family

引用

IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES 2006年第8期E89A卷 2140-2148页

作者： Jansen, Boris Nakayama, Kenji Kanazawa Univ Grad Sch Nat Sci & Technol Kanazawa Ishikawa 9201192 Japan

Over the years, many improvements and refinements to the backpropagation learning algorithm have been reported. In this paper, a new adaptive penalty-based learning extension for the backpropagation learning algorithm and its variants is proposed. The new method initially puts pressure on artificial neural networks in order to get all outputs for all training patterns into the correct half of the output range, instead of mainly focusing on minimizing the difference between the target and actual output values. The upper bound of the penalty values is also controlled. The technique is easy to implement and computationally inexpensive. In this study, the new approach is applied to the backpropagation learning algorithm as well as the RPROP learning algorithm. The superiority of the new proposed method is demonstrated though many simulations. By applying the extension, the percentage of successful runs can be greatly increased and the average number of epochs to convergence can be well reduced on various problem instances. The behavior of the penalty values during training is also analyzed and their active role within the learning process is confirmed.

关键词： backpropagation learning algorithm convergence error function neural networks generalization

来源：评论

学校读者我要写书评

暂无评论

A learning rule for very simple universal approximators consisting of a single layer of perceptrons

引用

NEURAL NETWORKS 2008年第5期21卷 786-795页

作者： Auer, Peter Burgsteiner, Harald Maass, Wolfgang Graz Univ Appl Sci Dept Hlth Care Engn A-8020 Graz Austria Univ Min & Met Leoben Chair Informat Technol A-8700 Leoben Austria Graz Univ Technol Inst Theoret Comp Sci A-8010 Graz Austria

One may argue that the simplest type of neural networks beyond a single perceptron is an array of several perceptrons in parallel. In spite of their simplicity, Such circuits can compute any Boolean function if one views the majority of the binary perceptron outputs as the binary Output Of the parallel perceptron, and they are universal approximators for arbitrary continuous functions With values in 10, 11 if one views the fraction of perceptrons that Output I as the analog output of the parallel perceptron. Note that in contrast to the familiar model of a "multi-layer perceptron" the parallel perceptron that we consider here has just binary values as outputs of gates on the hidden layer. For a long time one has thought that there exists no competitive learning algorithm for these extremely simple neural networks, which also came to be known as committee machines. It is commonly assumed that one has to replace the hard threshold gates on the hidden layer by sigmoidal gates (or RBF-gates) and that one has to tune the weights on at least two successive layers in order to achieve satisfactory learning results for any class of neural networks that yield universal approximators. We show that this assumption is not true, by exhibiting a simple learning algorithm for parallel perceptrons - the parallel delta ride (p-delta rule). In contrast to backprop for multi-layer perceptrons, the p-delta rule only has to tune a single layer of weights, and it does not require the computation and Communication of analog values with high precision. Reduced communication also distinguishes our new learning rule from other learning rules for parallel perceptrons Such as MADALINE. Obviously these features make the p-delta rule attractive as a biologically more realistic alternative to backprop in biological neural circuits, but also for implementations in special purpose hardware. We show that the p-delta rule also implements gradient descent - with regard to a suitable error measure - alt

关键词： learning algorithm parallel delta rule parallel perceptrons committee machines reduced communication

来源：评论

学校读者我要写书评

暂无评论

An Improved Alpha Beta Filter Using a Deep Extreme learning Machine

引用

IEEE ACCESS 2021年 9卷 61548-61564页

作者： Khan, Junaid Fayaz, Muhammad Hussain, Ayyaz Khalid, Shah Mashwani, Wali Khan Gwak, Jeonghwan Sungkyunkwan Univ Intelligent Syst Res Inst Coll Informat & Commun Engn Suwon 440746 South Korea Univ Cent Asia Dept Comp Sci Naryn 722918 Kyrgyzstan Quaid I Azam Univ Dept Comp Sci Islamabad 45320 Pakistan Univ Malakand Dept Comp Sci & IT Chakdara 18800 Pakistan Kohat Univ Sci & Technol Inst Numer Sci Kohat 26000 Pakistan Korea Natl Univ Transportat Dept Software Chungju 27469 South Korea Korea Natl Univ Transportat Dept Biomed Engn Chungju 27469 South Korea Korea Natl Univ Transportat Dept AI Robot Engn Chungju 27469 South Korea Korea Natl Univ Transportat Dept IT & Energy Convergence BK21 FOUR Chungju 27469 South Korea

This paper introduces new learning to the prediction model to enhance the prediction algorithms' performance in dynamic circumstances. We have proposed a novel technique based on the alpha-beta filter and deep extreme learning machine (DELM) algorithm named as learning to alpha-beta filter. The proposed method has two main components, namely the prediction unit and the learning unit. We have used the alpha-beta filter in the prediction unit, and the learning unit uses a DELM. The main problem with the conventional alpha-beta filter is that the values are generally selected via the trial-and-error technique. Once the alpha-beta values are chosen for a specific problem, they remain fixed for the entire data. It has been observed that different alpha-beta values for the same problem give different results. Hence it is essential to tune the alpha-beta values according to their historical behavior for certain values. Therefore, in the proposed method, we have addressed this problem and added the learning module to the conventional alpha-beta filter toimprove the alpha-beta filter's performance. The DELM algorithm has been used to enhance the conventional alpha-beta filter algorithm's performance in dynamically changing conditions. The model performance has been measured using indoor environmental values of temperature and humidity. The relative improvement in the proposed learning prediction model's accuracy was 7.72% and 16.47% in RMSE and MSE metrics. The results show that the proposed model outperforms in terms of the result as compared to the conventional alpha-beta filter.

关键词： Alpha-beta filter learning algorithm prediction algorithm deep extreme learning machine energy prediction

来源：评论

学校读者我要写书评

暂无评论

THE COMPLEXITY OF learning IN PLN NETWORKS

引用

NEURAL NETWORKS 1995年第2期8卷 221-228页

作者： ZHANG, B ZHANG, LN ZHANG, H ANHUI UNIV HEFEI 230039PEOPLES R CHINA

In this paper, the complexity of learning in the feedforward PLN network is investigated by using Markov chain theory, when its training samples are incomplete (i.e., a network with hidden nodes). We present a learning algorithm. A formula for computing the average number of steps that the learning algorithm converges is obtained when the PLN network exists a solution. In the probabilistic sense, the completeness of the learning algorithm is proved. Some computer simulations are given to verify the analysis.

关键词： learning algorithm MARKOV CHAINS TRANSITION MATRIX ABSORBING STATE STABLE STATE PLN NETWORK

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：