FEATURE SELECTION BASED ON COMPACTNESS AND SEPARABILITY: COMPARISON WITH FILTER‐BASED METHODS

Selecting a subset of salient features for performing clustering using a clustering learning algorithm has been explored extensively in many real‐world applications. To select salient features during training, the filter model evaluates the intrinsic characteristics of each individual feature but is not permitted to use a clustering learning algorithm that provides clustered information to train the features. In particular, the filter model aims to predict unobservable clusters and measure how the features help provide satisfactory within‐cluster and between‐cluster scatters to achieve a good clustering quality. However, it is generally difficult to achieve both scatters in the filter model. For example, a random variable with a large variance may raise only the between‐cluster scatter, whereas another variable following a uniform distribution may raise only the within‐cluster scatter. In this paper, we present a new filter‐based method to quantify features that consider feature compactness and separability to ensure that both scatters are raised. Moreover, our method adopts a new search strategy to locate the best feature salience vector instead of visiting the space of all the possible feature subsets. After the benchmark data sets are tested, the experimental results indicate that our method performs better than many benchmark filter‐based methods at selecting a feature subset to perform clustering.     

多目标优化在特征选择子集评价中的应用

特征选择是处理高维大数据常用的降维手段,但其中牵涉到的多个彼此冲突的特征子集评价目标难以平衡。为综合考虑特征选择中多种子集评价方式间的折中,优化子集性能,提出一种基于子集评价多目标优化的特征选择框架,并重点对多目标粒子群优化（MOPSO）在特征子集评价中的应用进行了研究。该框架分别根据子集的稀疏度、分类能力和信息损失度设计多目标优化函数,继而基于多目标优化算法进行特征权值向量寻优,并通过权值向量Pareto解集膝点选取确定最优向量,最终实现基于权值向量排序的特征选择。设计实验对比了基于多目标粒子群优化算法的特征选择（FS_MOPSO）与四种经典方法的性能,多个数据集上的结果表明,FS_MOPSO在低维空间表现出更高的分类精度,并保证了更少的信息损失。     

An improvement on floating search algorithms for feature subset selection

A new improved forward floating selection (IFFS) algorithm for selecting a subset of features is presented. Our proposed algorithm improves the state-of-the-art sequential forward floating selection algorithm. The improvement is to add an additional search step called “replacing the weak feature” to check whether removing any feature in the currently selected feature subset and adding a new one at each sequential step can improve the current feature subset. Our method provides the optimal or quasi-optimal (close to optimal) solutions for many selected subsets and requires significantly less computational load than optimal feature selection algorithms. Our experimental results for four different databases demonstrate that our algorithm consistently selects better subsets than other suboptimal feature selection algorithms do, especially when the original number of features of the database is large.     

面向高维数据的安全半监督分类算法

半监督学习过程中,由于无标记样本的随机选择造成分类器性能降低及不稳定性的情况经常发生;同时,面对仅包含少量有标记样本的高维数据的分类问题,传统的半监督学习算法效果不是很理想.为了解决这些问题,本文从探索数据样本空间和特征空间两个角度出发,提出一种结合随机子空间技术和集成技术的安全半监督学习算法（A safe semi-supervised learning algorithm combining stochastic subspace technology and ensemble technology,S3LSE）,处理仅包含极少量有标记样本的高维数据分类问题.首先,S3LSE采用随机子空间技术将高维数据集分解为B个特征子集,并根据样本间的隐含信息对每个特征子集优化,形成B个最优特征子集;接着,将每个最优特征子集抽样形成G个样本子集,在每个样本子集中使用安全的样本标记方法扩充有标记样本,生成G个分类器,并对G个分类器进行集成;然后,对B个最优特征子集生成的B个集成分类器再次进行集成,实现高维数据的分类.最后,使用高维数据集模拟半监督学习过程进行实验,实验结果表明S3LSE具有较好的性能.     

A genetic algorithm-based method for feature subset selection

As a commonly used technique in data preprocessing, feature selection selects a subset of informative attributes or variables to build models describing data. By removing redundant and irrelevant or noise features, feature selection can improve the predictive accuracy and the comprehensibility of the predictors or classifiers. Many feature selection algorithms with different selection criteria has been introduced by researchers. However, it is discovered that no single criterion is best for all applications. In this paper, we propose a framework based on a genetic algorithm (GA) for feature subset selection that combines various existing feature selection methods. The advantages of this approach include the ability to accommodate multiple feature selection criteria and find small subsets of features that perform well for a particular inductive learning algorithm of interest to build the classifier. We conducted experiments using three data sets and three existing feature selection methods. The experimental results demonstrate that our approach is a robust and effective approach to find subsets of features with higher classification accuracy and/or smaller size compared to each individual feature selection algorithm.     

融合序列后向选择与支持向量机的混合式特征选择算法

维度灾难是机器学习任务中的常见问题,特征选择算法能够从原始数据集中选取出最优特征子集,降低特征维度.提出一种混合式特征选择算法,首先用卡方检验和过滤式方法选择重要特征子集并进行标准化缩放,再用序列后向选择算法（SBS）与支持向量机（SVM）包裹的SBS-SVM算法选择最优特征子集,实现分类性能最大化并有效降低特征数量.实验中,将包裹阶段的SBS-SVM与其他两种算法在3个经典数据集上进行测试,结果表明,SBS-SVM算法在分类性能和泛化能力方面均具有较好的表现.     

使用EBIC的软件故障特征选择方法

软件故障预测中若采用大量度量指标建立预测模型,可能因其中含有无关特征使预测模型性能受到不良影响,故障预测中的特征选择步骤选取一定维度的部分故障数据建立预测模型来提高模型性能,以达到压缩特征维度,提高模型预测精度,降低预测模型复杂度,节约计算资源的目的。传统特征排序方法仅评估单个特征对类标的影响,建立的预测模型有效性较低;特征子集选择方法需搜索所有特征子集,耗费计算资源且所选特征维数较高。针对以上问题,提出一种基于拓展贝叶斯信息准则的特征选择方法(EBIC-FS),该方法对数据进行线性回归,并计算出残差平方和较小且数据维数较少的特征模型。在公开数据集M&R及Promise上进行实验,结果表明该方法能有效压缩特征维度,且预测模型性能与5种基线方法相比有较大提升。     

A novel hybrid algorithm for feature selection

Feature selection is an important filtering method for data analysis, pattern classification, data mining, and so on. Feature selection reduces the number of features by removing irrelevant and redundant data. In this paper, we propose a hybrid filter–wrapper feature subset selection algorithm called the maximum Spearman minimum covariance cuckoo search (MSMCCS). First, based on Spearman and covariance, a filter algorithm is proposed called maximum Spearman minimum covariance (MSMC). Second, three parameters are proposed in MSMC to adjust the weights of the correlation and redundancy, improve the relevance of feature subsets, and reduce the redundancy. Third, in the improved cuckoo search algorithm, a weighted combination strategy is used to select candidate feature subsets, a crossover mutation concept is used to adjust the candidate feature subsets, and finally, the filtered features are selected into optimal feature subsets. Therefore, the MSMCCS combines the efficiency of filters with the greater accuracy of wrappers. Experimental results on eight common data sets from the University of California at Irvine Machine Learning Repository showed that the MSMCCS algorithm had better classification accuracy than the seven wrapper methods, the one filter method, and the two hybrid methods. Furthermore, the proposed algorithm achieved preferable performance on the Wilcoxon signed-rank test and the sensitivity–specificity test.     

混合式的K-匿名特征选择算法

K-匿名算法通过对数据的泛化、隐藏等手段使得数据达到K-匿名条件,在隐藏特征的同时考虑数据的隐私性与分类性能,可以视为一种特殊的特征选择方法,即K-匿名特征选择。K-匿名特征选择方法结合K-匿名与特征选择的特点使用多个评价准则选出K-匿名特征子集。过滤式K-匿名特征选择方法难以搜索到所有满足K-匿名条件的候选特征子集,不能保证得到的特征子集的分类性能最优,而封装式特征选择方法计算成本很大,因此,结合过滤式特征排序与封装式特征选择的特点,改进已有方法中的前向搜索策略,设计了一种混合式K-匿名特征选择算法,使用分类性能作为评价准则选出分类性能最好的K-匿名特征子集。在多个公开数据集上进行实验,结果表明,所提算法在分类性能上可以超过现有算法并且信息损失更小。     

Feature Subset Selection within a Simulated Annealing Data Mining Algorithm

An overview of the principle feature subset selection methods isgiven. We investigate a number of measures of feature subset quality, usinglarge commercial databases. We develop an entropic measure, based upon theinformation gain approach used within ID3 and C4.5 to build trees, which isshown to give the best performance over our databases. This measure is usedwithin a simple feature subset selection algorithm and the technique is usedto generate subsets of high quality features from the databases. A simulatedannealing based data mining technique is presented and applied to thedatabases. The performance using all features is compared to that achievedusing the subset selected by our algorithm. We show that a substantialreduction in the number of features may be achieved together with animprovement in the performance of our data mining system. We also present amodification of the data mining algorithm, which allows it to simultaneouslysearch for promising feature subsets and high quality rules. The effect ofvarying the generality level of the desired pattern is alsoinvestigated.