GENETIC ALGORITHMS AND ARTIFICIAL NEURAL NETWORK(Ⅱ)──TRAINING ARTIFICIAL NEURAL NETWORK BY EGA-GDR

您当前的位置：

首页 >

文章列表页 >

更新时间：2025-12-16

- GENETIC ALGORITHMS AND ARTIFICIAL NEURAL NETWORK(Ⅱ)──TRAINING ARTIFICIAL NEURAL NETWORK BY EGA-GDR
- CIESC Journal Issue 4, (1996)
- 作者机构：
  
  清华大学化工系!北京100084,清华大学化工系!北京100084,清华大学化工系,北京,100084
- 作者简介：
- 基金信息：
- DOI：
  CLC： 224
- Online First：25 August 1996，
  
  Published：25 August 1996
- 稿件说明：
移动端阅览
Chen Fangze, Chen Bingzhen, He Xiaorong（Department of Chemical Engineering. GENETIC ALGORITHMS AND ARTIFICIAL NEURAL NETWORK(Ⅱ)──TRAINING ARTIFICIAL NEURAL NETWORK BY EGA-GDR[J]. CIESC Journal, 1996, (4).
DOI：

Chen Fangze, Chen Bingzhen, He Xiaorong（Department of Chemical Engineering. GENETIC ALGORITHMS AND ARTIFICIAL NEURAL NETWORK(Ⅱ)──TRAINING ARTIFICIAL NEURAL NETWORK BY EGA-GDR[J]. CIESC Journal, 1996, (4). DOI：

为了更有效地解决神经网络培训中遇到的局部最优解问题，提出一种融合EGA和GDR算法特长的新算法──EGA-GDR算法．通过分别对拓扑为2×3×3和8×8×4两个神经网络的训练，表明该算法较之SA-GDR算法可以更有效地解决神经网络训练中的局部最优解问题．

In order to overcome a local optimal solution which often occurs in the training process of ANN

a new algorithm

called EGA-GDR

is proposed

which combines the perfect characters of both EGA and GDR.The results obtained from training two ANNs whose structures are2×3×1 and8×8×4 respectively

show that EGA-GDR can more effectively search the optimal solution than SA—GDR does.

Views

482

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Performance study of gas-liquid vortex separators under low cross-sectional gas velocity

Exothermic characteristics of energy carrier for calcium looping thermo-chemical energy storage

Hydrolysis reaction mechanism of CNCl on the surface of composite metal CuCr2O4

Construction of transition metal-based nitrogen-doped carbon and their desulfurization performance

Research progress of ionic liquids in separation of aromatics/alkanes and olefins/alkanes

Related Author

No data

Related Institution

No data

⁰