偏最小二乘判别分析结合F-score用于蛋白质组学质谱数据的特征筛选(英文)

提出了一种基于偏最小二乘判别分析和F-score的特征筛选方法,并将其用于蛋白质组学质谱数据分析。该方法主要包含3个步骤:(1)用LIMPIC算法对原始数据进行预处理;(2)计算每个变量的F-score值并将所有变量按F-score值降底的顺序排列;(3)采用偏最小二乘判别分析交互检验按前向选择法选择最佳变量子集。用本方法对一组结肠癌数据进行分析,最终从原始的16331个质荷比变量中选择了8个特征质荷比作为潜在的生物标记物。用所选择的特征对独立测试集的样本进行判别,其灵敏度和特异性分别达到了95.24%和100%。结果表明,所提出的方法可用于蛋白质组学质谱数据的特征筛选及样本分类。     

最小冗余最大分离准则特征选择方法

特征选择是处理高维数据的一项有效技术。针对传统方法的不足,结合[F-score]与互信息,提出了一种最小冗余最大分离的特征选择评价准则,该准则使所选择的特征具有更好的分类和预测能力;采用二进制布谷鸟搜索算法和二次规划两种搜索策略来搜索最优特征子集,并对两种搜索策略的准确性和计算量进行分析比较;最后,利用UCI数据集进行实验测试,实验结果说明了所提理论的有效性。     

Diversity-aware classifier ensemble selection via f-score

The primary effect of using a reduced number of classifiers is a reduction in the computational requirements during learning and classification time. In addition to this obvious result, research shows that the fusion of all available classifiers is not a guarantee of best performance but good results on the average. The much researched issue of whether it is more convenient to fuse or to select has become even more of interest in recent years with the development of the Online Boosting theory, where a limited set of classifiers is continuously updated as new inputs are observed and classifications performed. The concept of online classification has recently received significant interest in the computer vision community. Classifiers can be trained on the visual features of a target, casting the tracking problem into a binary classification one: distinguishing the target from the background.Here we discuss how to optimize the performance of a classifier ensemble employed for target tracking in video sequences. In particular, we propose the F-score measure as a novel means to select the members of the ensemble in a dynamic fashion. For each frame, the ensemble is built as a subset of a larger pool of classifiers selecting its members according to their F-score. We observed an overall increase in classification accuracy and a general tendency in redundancy reduction among the members of an f-score optimized ensemble. We carried out our experiments both on benchmark binary datasets and standard video sequences.     

基于改进的F-score与支持向量机的特征选择方法

将传统F-score度量样本特征在两类之间的辨别能力进行推广,提出了改进的F-score,使其不但能够评价样本特征在两类之间的辨别能力,而且能够度量样本特征在多类之间的辨别能力大小。以改进的F-score作为特征选择准则,用支持向量机(SVM)评估所选特征子集的有效性,实现有效的特征选择。通过UCI机器学习数据库中六组数据集的实验测试,并与SVM、PCA+SVM方法进行比较,证明基于改进F-score与SVM的特征选择方法不仅提高了分类精度,并具有很好的泛化能力,且在训练时间上优于PCA+SVM方法。     

改进F-score特征选择的MPSO-BP神经网络短期负荷预测

针对海量数据提出一种基于改进Fisher分数(F-score)特征选择的改进粒子群优化的BP(Modified Particle Swarm Optimization and Back Propagation,MPSO-BP)神经网络短期负荷预测方法。首先采用改进F-score特征评价准则计算影响负荷预测精度各个特征的F-score值,再通过F-score Area法设定阈值筛选出最优特征子集,然后将最优特征子集作为MPSO-BP神经网络模型的输入变量完成对预测日一天24点负荷的预测,并与MPSO-BP神经网络短期负荷预测和传统BP神经网络短期负荷预测进行对比。算例表明,文中提出的短期负荷预测方法可以较好地对海量数据进行挖掘,具有较高的预测精度。     

改进的F-score算法在语音情感识别中的应用

针对F-score特征选择算法不能揭示特征间互信息而不能有效降维这一问题,应用去相关的方法对F-score进行改进,利用德语情感语音库EMO-DB,在提取语音情感特征的基础上,根据支持向量机（SVM）的分类精度选择出分类效果最佳的特征子集。与F-score特征选择算法对比,改进后的算法实现了候选特征集较大幅度的降维,选择出了有效的特征子集,同时得到了较理想的语音情感识别效果。     

Classification of brain signal (EEG) induced by shape-analogous letter perception

Visual perception of English letters involves different underlying brain processes including brain activity alteration in multiple frequency bands. However, shape analogous letters elicit brain activities which are not obviously distinct and it is therefore difficult to differentiate those activities. In order to address discriminative feasibility and classification performance of the perception of shape-analogous letters, we performed an experiment in where EEG signals were obtained from 20 subjects while they were perceiving shape analogous letters (i.e., ‘p’, ‘q’, ‘b’, and ‘d’). Spectral power densities from five typical frequency bands (i.e., delta, theta, alpha, beta and gamma) were extracted as features, which were then classified by either individual widely-used classifiers, namely k-Nearest Neighbors (kNN), Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), Random Forest (RF) and AdaBoost (ADA), or an ensemble of some of them. The F-score was employed to select most discriminative features so that the dimension of features was reduced. The results showed that the RF achieved the highest accuracy of 74.1% in the case of multi-class classification. In the case of binary classification, the best performance (Accuracy 86.39%) was achieved by the RF classifier in terms of average accuracy across all possible pairs of the letters. In addition, we employed decision fusion strategy to exert complementary strengths of different classifiers. The results demonstrated that the performance was elevated from 74.10% to 76.63% for the multi-class classification and from 86.39% to 88.08% for the binary class classification.