This paper mainly concerns two kinds of approaches : nonlinear unknown input observer scheme and the method based on adaptive learning , and summarizes some other approaches 本文着重讨论了两类方法：非线性未知输入观测器方法和基于自适应学习的方法，并简要概述了其它一些方法。
Intelligent e - knowledge base engine for any kind of knowledge - based applications , supporting effective knowledge presentation , precise knowledge search , adaptive learning and immediate consulting 智能电子知识库引擎供任意基于知识的应用程序使用，支持有效的知识表达，精确的知识搜索，适应性学习和即时查询。
General knowledge machine project - intelligent e - knowledge base engine for any kind of knowledge - based applications , supporting effective knowledge presentation , precise knowledge search , adaptive learning and immediate consulting 智能电子知识库引擎供任意基于知识的应用程序使用，支持有效的知识表达，精确的知识搜索，适应性学习和即时查询。
The sliding mode variable - structure controller basing attitude error quaternion and the sliding mode variable - structure controller basing adaptive learning on rbf neural network , three learning arithmetics of rbf neural network is emphasized 针对卫星平台分别设计了基于姿态误差四元数的滑模变结构控制器和基于rbf神经网络上界自适应学习的滑模控制器。
In view of adaptive learning for diagnostic images recognition , studying the theorem and the rule of art artificial neural network to learning and adaptive , provided an algorithm with a function of self - stability and classifier learning 摘要针对特征图像的自适应识别技术问题，研究了利用具有学习和自适应功能的art人工神经网络原理及规则，提出了具有自稳定适应和分类学习功能的算法。
Aiming at the shortcomings of bp model such as easy leading to local minimum value and slow convergence velocity , some improved methods are presented such as non - linear normalization of original datax additional momentum and automatically adaptive learning rate 郑州大学硕士学位论文中/英文摘要( 5 )选取河南省淮阳市为研究对象，建立了该地区的需水量预测模型和水质一水量联合优化配置模型。
This method effectively shield the learning process from plenty of feedback noises produced by distorted threshold and sparseness of initial information , and also imitate artificial feedback approximately to perfect the intelligence of adaptive learning mechanism 此外,选择聚类择优后的一类信息参与rocchio自适应学习与更新,能够有效地削弱阈值偏见性和伪相关反馈排序偏见性造成的负面影响。
Different training algorithms , namely levenburg - marquart algorithm and the gradient - based algorithm with an adaptive learning rate and momentum , are compared in this paper . according to the engineering requirement , dimensions of idc can be designed using the trained ann model and ga 将自适应调整学习率并加入动量因子的梯度下降法和levengurg - marquart训练算法的训练结果做了比较分析，同时引入了性能函数的改进形式。
4 . when the regular bp neural network is used , it may have poor performance because of bad convergence . so we choose improved bp neural network as classifier , which combines momentum and adaptive learning rate , and improve the learning and recognition efficiency 针对标准的bp神经网络常因收敛速度太慢而影响求解质量的问题，我们选用动量法和学习率自适应调整相结合的改进的bp神经网络作为分类器，从而提高了学习速度和识别率。
I introduction : describes the disadvantages of learning system based on web . then brings forward that it is necessary to design and develop adaptive learning system based on web , and claim that adaptive learning system should provide with five basic functions 全文分为六部分，具体安排如下：第一部分引言：分析当前基于web的学习系统存在的弊端，提出了开发基于web的适应性学习系统的必要性，概括了基于web适应性学习系统应该具有的基本功能。
