| 一种不平衡注意参数条件下的遗传协同学习算法 |
| |
| 引用本文： | 王海龙,戚飞虎,詹劲峰.一种不平衡注意参数条件下的遗传协同学习算法[J].电子学报,2000,28(11):25-28. |
| |
| 作者姓名： | 王海龙 戚飞虎 詹劲峰 |
| |
| 作者单位： | 上海交通大学计算机科学与工程系,上海 200030 |
| |
| 基金项目： | 国家自然科学基金! (No .69772 0 0 2 ) |
| |
| 摘 要： | 本文讨论了目标识别的协同方法在不平衡注意参数条件下的动力学行为,并提出了不平衡注意参数条件下的遗传协同学习算法(GSLA).该算法利用遗传算法的全局最优搜索能力,对协同神经网络的注意参数进行全局优化.对从"车牌识别系统"中得到的数字样本的实验证明:新算法能有效地在注意参数空间搜索全局最优解,挖掘出协同方法在目标识别方面的最大潜能.另外,本文还将新算法与利用奖惩学习算法的协同学习算法进行了全局优化能力的比较,发现新算法具有收敛快和全局最优搜索能力强的特点.
|
| 关 键 词： | 注意参数 目标识别 协同神经网络 神经网络优化 遗传算法 奖惩学习算法 |
| 文章编号： | 0372-2112(2000)11-0025-04 |
| 收稿时间： | 1999-09-05 |
A Genetic-synergetic Learning Algorithm under Unbalanced Attention Parameters |
| |
| WANG Hai-long,QI Fei-hu,ZHAN Jin-feng.A Genetic-synergetic Learning Algorithm under Unbalanced Attention Parameters[J].Acta Electronica Sinica,2000,28(11):25-28. |
| |
| Authors: | WANG Hai-long QI Fei-hu ZHAN Jin-feng |
| |
| Affiliation: | Computer Science and Engineering Department of Shanghai JiaoTong University,Shanghai 200030,China |
| |
| Abstract: | An analysis of dynamics of synergetic approach on target recognition under unbalanced attention parameters is presented and a genetic synergetic learning algorithm (GSLA) for synergetic neural network (SNN) is proposed in this paper.The GSLA is demonstrated to have strong ability of searching globally optimal solutions in the space of attention parameters,and makes the SNN do its best on target recognition.In addition,A comparison between the GSLA and the synergetic learning algorithm based on award penalty learning mechanism is made,and shows that the GSLA can converge more quickly and has stronger power of globally optimal searching than the algorithm based on award penalty learning mechanism. |
| |
| Keywords: | attention parameters target recognition synergetic neural networks optimization of neural networks genetic algorithm award penalty learning mechanism |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《电子学报》浏览原始摘要信息 |
|
点击此处可从《电子学报》下载全文 |
