Learning NMF Representation Using a Hybrid Method Combining Feasible Direction Algorithm and Simulated Annealing

CHEN Wei-gang; QI Fei-hu

您当前的位置：

首页 >

文章列表页 >

Learning NMF Representation Using a Hybrid Method Combining Feasible Direction Algorithm and Simulated Annealing

更新时间：2025-07-16

- Learning NMF Representation Using a Hybrid Method Combining Feasible Direction Algorithm and Simulated Annealing
- Acta Electronica Sinica Vol. 31, Issue S1, Pages: 2190-2193(2003)
- 作者机构：
  
  上海交通大学计算机科学与工程系,上海,200030
- 作者简介：
- 基金信息：
- DOI：
  CLC： TP391
- Published Online：25 December 2003，
  
  Published：2003
- 稿件说明：
移动端阅览
CHEN Wei-gang, QI Fei-hu. Learning NMF Representation Using a Hybrid Method Combining Feasible Direction Algorithm and Simulated Annealing[J]. Acta Electronica Sinica, 2003, 31(S1): 2190-2193.
DOI：

CHEN Wei-gang, QI Fei-hu. Learning NMF Representation Using a Hybrid Method Combining Feasible Direction Algorithm and Simulated Annealing[J]. Acta Electronica Sinica, 2003, 31(S1): 2190-2193. DOI：

摘要

NMF子空间特征提取被表示成一个大规模线性约束非线性优化问题.为了获得更优性能的基图像

设计了一个可行方向算法结合模拟退火算法的混合算法来求解这个优化问题.以基于梯度的可行方向算法作为局部寻优的手段

加快收敛速度;以模拟退火算法作为全局寻优的手段

避免优化过程陷入局部极小点.同时

在模拟退火操作中

采用对比度增强算法

使获得的基图像更加地空间局部化.实验表明

本文的可行方向算法比采用归一化实现等式约束的原算法在学习的最后阶段有更好的收敛速度

所获得的基图像更加地空间局部化

而且在人脸重建的应用中有更好的性能.

Abstract

Non-negative matrix factorization(NMF) is formulated as a large-scale optimization problem with linear equality constraints.To get better performance

a hybrid combining simulated annealing(SA) and gradient-based algorithm is designed.The simulated annealing gradually produces better solutions with the gradient-based algorithm serving as an accelerator.Experimental results are presented to compare our method and the original method for learning NMF representation

which demonstrate the proposed method can learn basis images more spatially localized and perform better than the original one at later training stages.The comparison study on face reconstruction also shows that the proposed method leads to better results than PCA and the original NMF.

关键词

Keywords

references

Views

1215

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

A Salp Swarm Algorithm Based on Stepped Tent Chaos and Simulated Annealing

Build Kd-Tree for Virtual Scenes in a Fast and Optimal Way

Hybrid Artificial Fish Swarm Optimization Algorithm Based on Mutation Operator and Simulated Annealing

Compounding Reconstruction for Synthetic Aperture Radar Images

Speckle Removal for SAR Single-look Images by Simulated Annealing

Related Author

DONG Chao-yi

CHEN Xiao-yan

QI Yong-sheng

ZHAO Xiao-yi

WANG Qi-lai

ZHOU Peng

PAN Jin-gui

XU Xiao-yang

Related Institution

Inner Mongolia Electromechanical Control Laboratory

Inner Mongolia University of Technology

State Key Laboratory for Novel Software Technology, Nanjing University

School of Electronic and Information EngineeringDalian University of TechnologyDalianLiaoning 116024China

Zhengzhou Institute of Light IndustryZhengzhouHenan 450002China

⁰