检索结果-内蒙古大学图书馆

3rd World Congress on Intelligent Control and Automation

作者： Li, DM Liu, JQ Hu, HZ Harbin Inst Technol Harbin 150001 Peoples R China

ISBN: (纸本)7312012035

A novel fast algorithm for traing a class of fuzzy neural networks (FNN) is studied. The proposed algorithm is named as Least Square-Simplex (LS-Simplex). The algorithm obtains the performance of global convergence and avoids the inherent local convergence when adopting grads algorithm to train FNN. Also it accelerates FNN's training and can be used on-line which is impossible when using genetic algorithm (GA). Compared with grads algorithm and GA, the LS-Simplex owns more accurate precision and faster convergent speed, and the FNN obtained has excellent performance of generalization.

关键词： fuzzy neural networks training algorithm least square technique simplex search algorithm

来源：评论

学校读者我要写书评

暂无评论

Improved prediction accuracy of biomass heating value using proximate analysis with various ANN training algorithms

引用

RESULTS IN ENGINEERING 2022年 16卷

作者： Veza, Ibham Irianto Panchal, Hitesh Paristiawan, Permana Andi Idris, Muhammad Fattah, I. M. Rizwanul Putra, Nicky R. Silambarasan, Rajendran Univ Teknol PETRONAS Dept Mech Engn Bandar Seri Iskandar 32610 Perak Darul Rid Malaysia Rabdan Acad Fac Resilence Dept Gen Educ Abu Dhabi U Arab Emirates Govt Engn Coll Dept Mech Engn Patan Gujarat India Natl Res & Innovat Agcy Res Ctr Met South Tangerang 15314 Banten Indonesia PT PLN Persero Engn & Technol Div Jakarta Indonesia Univ Technol Sydney Fac Engn & IT Ctr Technol Water & Wastewater Sch Civil & Environm Engn Ultimo NSW 2007 Australia Univ Teknol Malaysia Ctr Lipid Engn & Appl Res CLEAR Ibnu Sina Inst Sci & Ind Res Johor Baharu 81310 Malaysia JKK Nattraja Coll Engn & Technol Dept Mech Al Engn Namakkal India

The conventional experimental methods to determine biomass heating value are laborious and costly. Numerous correlations to estimate biomass' higher heating values have been proposed using proximate analysis. Recently, the utilisation of artificial neural network (ANN) has been extensively applied to predict HHV. However, most studies of ANN to estimate the biomass' HHV only use one algorithm to train a small number of biomass datasets. The specific objective of this study is to predict the HHV of 350 samples of biomass from the proximate analysis by developing an ANN model which was trained with 11 different algorithms. This study fills a gap in the research on how to predict the HHV of biomass using numerous ANN training algorithms utilising sizeable biomass datasets. Results show that the ANN trained with Levenberg-Marquardt gave the highest accuracy. The Levenberg-Marquardt algorithm shows the best fit giving the highest R and R-2 values and the lowest MAD, MSE, RMSE and MAPE. Compared with previous biomass HHV prediction studies, the ANN model developed in this study provides improved prediction accuracy with higher R2 and lower RMSE. Results from this study have also indicated that the Levenberg-Marquardt should be the first-choice supervised algorithm for feedforward-backpropagation.

关键词： Biomass Higher heating value Artificial neural network ANN Levenberg-marquardt training algorithm

来源：评论

学校读者我要写书评

暂无评论

Application of different training methodologies for the development of a back propagation artificial neural network retention model in ion chromatography

引用

JOURNAL OF CHEMOMETRICS 2008年第1-2期22卷 106-113页

作者： Bolanca, Tornislav Cerjan-Stefanovic, Stefica Ukic, Sime Rogosic, Marko Lusa, Melita Univ Zagreb Fac Chem Engn & Technol Zagreb 10000 Croatia

The reliability of predicted separations in ion chromatography depends mainly on the accuracy of retention predictions. Any model able to improve this accuracy will yield predicted optimal separations closer to the reality. In this work artificial neural networks were used for retention modeling of void peak, fluoride, chlorite, chloride, chlorate, nitrate and sulfate. In order to increase performance characteristics of the developed model, different training methodologies were applied and discussed. Furthermore, the number of neurons in hidden layer, activation function and number of experimental data used for building the model were optimized in terms of decreasing the experimental effort without disruption of performance characteristics. This resulted in the superior predictive ability of developed retention model (average of relative error is 0.4533%). Copyright (c) 2008 John Wiley & Sons, Ltd.

关键词： artificial neural network training algorithm retention modeling ion chromatography

来源：评论

学校读者我要写书评

暂无评论

A fast training method for memristor crossbar based multi-layer neural networks

引用

ANALOG INTEGRATED CIRCUITS AND SIGNAL PROCESSING 2017年第3期93卷 443-454页

作者： Hasan, Raqibul Taha, Tarek M. Yakopcic, Chris Univ Dayton Dept Elect & Comp Engn Dayton OH 45469 USA

Memristor crossbar arrays carry out multiply-add operations in parallel in the analog domain which is the dominant operation in a neural network application. On-chip training of memristor neural network systems have the significant advantage of being able to get around device variability and faults. This paper presents a novel technique for on-chip training of multi-layer neural networks implemented using a single crossbar per layer and two memristors per synapse. Using two memristors per synapse provides double the synaptic weight precision when compared to a design that uses only one memristor per synapse. Proposed system utilizes a novel variant of the back-propagation (BP) algorithm to reduce both circuit area and training time. During training, all the memristors in a crossbar are updated in four steps in parallel. We evaluated the training of the proposed system with some nonlinearly separable datasets through detailed SPICE simulations which take crossbar wire resistance and sneak-paths into consideration. The proposed training algorithm trained the nonlinearly separable functions with a slight loss in accuracy compared to training with the traditional BP algorithm.

关键词： Neural networks Memristor crossbars training algorithm On-chip training

来源：评论

学校读者我要写书评

暂无评论

Comparative study of various training algorithms of artificial neural network

Comparative study of various training algorithms of artifici...

引用

International Conference on Advances in Computing, Communication Control and Networking

作者： Pragati Jaiswal Nikhil Kumar Gupta A. Ambikapathy Department of Information Technology Galgotias College of Engineering and Technology Greater Noida India Department of Electrical and Electronics Engineering Galgotias College of Engineering and Technology Greater Noida India

ISBN: (纸本)9781538641200

The idea of creating intelligent system has fascinated us with the advent of computer. The artificial neural network is a methodology for the development of intelligent system which can emulate as humans. Increasingly, new algorithms are researched for better efficiency and accuracy. Thereby, this paper examines the performance of different training algorithms for artificial neural network in the feedforward backpropagation architecture. The training algorithms are designed and executed on MATLAB R2017a for the analysis. The various training algorithms are analyzed to find the accuracy level between desired and actual results. The analysis is executed on house price prediction dataset from Kaggle.

关键词： Artificial neural networks training algorithm

来源：评论

学校读者我要写书评

暂无评论

A noise injection strategy for graph autoencoder training

引用

NEURAL COMPUTING & APPLICATIONS 2021年第10期33卷 4807-4814页

作者： Wang, Yingfeng Xu, Biyun Kwak, Myungjae Zeng, Xiaoqin Middle Georgia State Univ Dept Informat Technol Macon GA 31206 USA Beijing Kubao Technol Co Beijing 100124 Peoples R China Hohai Univ Inst Intelligence Sci & Technol Nanjing 210098 Jiangsu Peoples R China Univ Tennessee Dept Comp Sci & Engn Chattanooga TN 37403 USA

Graph autoencoder can map graph data into a low-dimensional space. It is a powerful graph embedding method applied in graph analytics to lower the computational cost. Researchers have developed different graph autoencoders for addressing different needs. This paper proposes a strategy based on noise injection for graph autoencoder training. This is a general training strategy that can flexibly fit most existing training algorithms. The experimental results verify this general strategy can significantly reduce overfitting and identify the noise rate setting for consistent training performance improvement.

关键词： Graph autoencoder Noise injection training algorithm Overfitting

来源：评论

学校读者我要写书评

暂无评论

A novel neural network training framework with data assimilation

引用

JOURNAL OF SUPERCOMPUTING 2022年第17期78卷 19020-19045页

作者： Chen, Chong Dou, Yixuan Chen, Jie Xue, Yaru China Univ Petr Coll Informat Sci & Engn Beijing 102249 Peoples R China

In recent years, the prosperity of deep learning has revolutionized the Artificial Neural Networks. However, the dependence of gradients and the offline training mechanism in the learning algorithms prevents the Artificial Neural Networks from further improvement. In this study, a gradient-free training framework based on data assimilation is proposed to avoid the calculation of gradients. In data assimilation algorithms, the error covariance between the forecasts and observations is used to optimize the states. The Feedforward Neural Networks are trained by gradient decent, data assimilation algorithms (Ensemble Kalman Filter and Ensemble Smoother with Multiple Data Assimilation), respectively. Ensemble Smoother with Multiple Data Assimilation trains Feedforward Neural Networks with pre-defined iterations by updating the parameters (i.e. states) using all the available observations which can be regarded as offline learning. Ensemble Kalman Filter optimizes Feedforward Neural Networks when new observation available by updating parameters which can be regarded as real-time learning. Two synthetic cases with the regression of a Sine function and a Mexican Hat function are conducted to validate the effectiveness of the proposed framework. Quantitative comparison with the root mean square error and coefficient of determination show that better performance is obtained by the proposed framework than the gradient decent method. Furthermore, the uncertainty of the parameters is quantified which shows the reduction in uncertainty along with the iterations in Ensemble Smoother with Multiple Data Assimilation. The proposed framework explores alternatives for real-time/offline training the existing Artificial Neural Networks (e.g. Convolutional Neural Networks, Recurrent Neural Networks) without the dependence of gradients and conducting uncertainty analysis at the same time.

关键词： Neural networks training algorithm Data assimilation EnKF ESMDA

来源：评论

学校读者我要写书评

暂无评论

MLPNN training via a Multiobjective Optimization of training Error and Stochastic Sensitivity

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2016年第5期27卷 978-992页

作者： Yeung, Daniel S. Li, Jin-Cheng Ng, Wing W. Y. Chan, Patrick P. K. S China Univ Technol Sch Comp Sci & Engn Guangzhou 510006 Guangdong Peoples R China Guangdong Pharmaceut Univ Coll Med Informat Engn Guangzhou 510006 Guangdong Peoples R China

The training of a multilayer perceptron neural network (MLPNN) concerns the selection of its architecture and the connection weights via the minimization of both the training error and a penalty term. Different penalty terms have been proposed to control the smoothness of the MLPNN for better generalization capability. However, controlling its smoothness using, for instance, the norm of weights or the Vapnik-Chervonenkis dimension cannot distinguish individual MLPNNs with the same number of free parameters or the same norm. In this paper, to enhance generalization capabilities, we propose a stochastic sensitivity measure (ST-SM) to realize a new penalty term for MLPNN training. The ST-SM determines the expectation of the squared output differences between the training samples and the unseen samples located within their Q-neighborhoods for a given MLPNN. It provides a direct measurement of the MLPNNs output fluctuations, i.e., smoothness. We adopt a two-phase Pareto-based multiobjective training algorithm for minimizing both the training error and the ST-SM as biobjective functions. Experiments on 20 UCI data sets show that the MLPNNs trained by the proposed algorithm yield better accuracies on testing data than several recent and classical MLPNN training methods.

关键词： Multilayer perceptron neural network (MLPNN) multiobjective sensitivity training algorithm

来源：评论

学校读者我要写书评

暂无评论

Model and training of QNN with weight

引用

NEURAL PROCESSING LETTERS 2006年第3期24卷 261-269页

作者： Zhou, Rigui Jiang, Nan Ding, Qiulin Nanjing Univ Aeronaut & Astronaut Dept Comp Sci & Technol Nanjing 210016 Jiangsu Peoples R China

Quantum Neural Network (QNN) is a burgeoning new field built upon the combination of classical neural networks and quantum computations, which has many problems needed to solve. Where the learning of the network weight vector is an issue must be settled to develop QNN. Upon the analysis of the Grover's quantum algorithm, a model of QNN with weight vector and a training method for it are proposed in this paper. It can be shown that this model and method work in quantum mechanism. Results on the data set show that this network model can deal with some classical problem such as XOR problem and the proposed weight updating algorithm based on the Grover always can learn training examples in a certain percentage, despiting it has not been proven to excel classical learning algorithm in performance. It yet has some advantages over classical counterpart.

关键词： grover algorithm QNN with weight training algorithm XOR problem

来源：评论

学校读者我要写书评

暂无评论

training TSVM with the proper number of positive samples

引用

PATTERN RECOGNITION LETTERS 2005年第14期26卷 2187-2194页

作者： Wang, Y Huang, ST Shanghai Jiao Tong Univ Comp Sci & Engn Dept Shanghai 200030 Peoples R China

The transductive support vector machine (TSVM) is the transductive inference of the support vector machine. The TSVM utilizes the information carried by the unlabeled samples for classification and acquires better classification performance than the regular support vector machine (SVM). As effective as the TSVM is, it still has obvious deficiency: The number of positive samples must be appointed before training and it is not changed during the training phase. This deficiency is caused by the pair-wise exchanging criterion used in the TSVM. In this paper, we propose a new transductive training algorithm by substituting the pair-wise exchanging criterion with the individually judging and changing criterion. Experimental results show that the new method releases the restriction of the appointment of the number of positive samples beforehand and improves the adaptability of the TSVM. (c) 2005 Elsevier B.V. All rights reserved.

关键词： transductive support vector machine training algorithm

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：