森林优化算法是一种基于森林中树木播种思想的演化算法,其具有良好的特征空间搜索能力,且实现难度低。但该算法在森林整体的收敛速度和寻优能力上仍存在提升空间,而且对高维数据集的适应度较差。本文针对上述问题提出了基于重复度分析...森林优化算法是一种基于森林中树木播种思想的演化算法,其具有良好的特征空间搜索能力,且实现难度低。但该算法在森林整体的收敛速度和寻优能力上仍存在提升空间,而且对高维数据集的适应度较差。本文针对上述问题提出了基于重复度分析的森林优化特征选择算法(feature selection using forest optimization algorithm based on duplication analysis, DAFSFOA)。该算法提出了基于信息增益的自适应初始化策略、森林重复度分析机制、森林重启机制、候选最优树生成策略、综合考虑特征选择数量和分类正确率的适应度函数。实验结果表明,DAFSFOA在大部分数据集上达到了最高的分类准确率。同时,对于高维数据集SRBCT,在维度缩减率和分类准确率方面,DAFSFOA对比森林优化特征选择算法(feature selection using forest optimization algorithm,FSFOA)都有较大提升。DAFSFOA比FSFOA具有更强的特征空间探索能力,而且能够适应不同维度的数据集。展开更多
Eigenvector subset selection is the key to face recognition. In this paper ,we propose ESS-BOA, a newrandomized, population-based evolutionary algorithm which deals with the Eigenvector Subset Selection (ESS)prob-lem ...Eigenvector subset selection is the key to face recognition. In this paper ,we propose ESS-BOA, a newrandomized, population-based evolutionary algorithm which deals with the Eigenvector Subset Selection (ESS)prob-lem on face recognition application. In ESS-BOA ,the ESS problem, stated as a search problem ,uses the BayesianOptimization Algorithm (BOA) as searching engine and the distance degree as the object function to select eigenvec-tor. Experimental results show that ESS-BOA outperforms the traditional the eigenface selection algorithm.展开更多
文摘森林优化算法是一种基于森林中树木播种思想的演化算法,其具有良好的特征空间搜索能力,且实现难度低。但该算法在森林整体的收敛速度和寻优能力上仍存在提升空间,而且对高维数据集的适应度较差。本文针对上述问题提出了基于重复度分析的森林优化特征选择算法(feature selection using forest optimization algorithm based on duplication analysis, DAFSFOA)。该算法提出了基于信息增益的自适应初始化策略、森林重复度分析机制、森林重启机制、候选最优树生成策略、综合考虑特征选择数量和分类正确率的适应度函数。实验结果表明,DAFSFOA在大部分数据集上达到了最高的分类准确率。同时,对于高维数据集SRBCT,在维度缩减率和分类准确率方面,DAFSFOA对比森林优化特征选择算法(feature selection using forest optimization algorithm,FSFOA)都有较大提升。DAFSFOA比FSFOA具有更强的特征空间探索能力,而且能够适应不同维度的数据集。
文摘Eigenvector subset selection is the key to face recognition. In this paper ,we propose ESS-BOA, a newrandomized, population-based evolutionary algorithm which deals with the Eigenvector Subset Selection (ESS)prob-lem on face recognition application. In ESS-BOA ,the ESS problem, stated as a search problem ,uses the BayesianOptimization Algorithm (BOA) as searching engine and the distance degree as the object function to select eigenvec-tor. Experimental results show that ESS-BOA outperforms the traditional the eigenface selection algorithm.