基于自适应邻域与聚类的非平衡数据特征选择

doi:10.3778/j.issn.1002-8331.2306-0248

摘要/Abstract

摘要： 为了解决传统邻域粗糙集未考虑不平衡数据的类分布，多数邻域系统通过人工调试难以找到最佳邻域半径，以及聚类时指定簇的数目等问题，提出一种基于自适应邻域与聚类的非平衡数据特征选择方法。根据样本在各个特征下与其他样本距离的平均值来确定样本的自适应[k]近邻和共享近邻，定义自适应邻域密度并设计混合采样模型，构建平衡决策系统。基于特征分布定义新的邻域半径，使用高斯核函数研究邻域内样本之间的模糊相似关系，使用模糊邻域互信息度量特征间的相关性，基于此对特征进行聚类。基于模糊邻域互信息构造粒子群初始化策略，并引入动态位掩码策略与适合整数编码的差异性扰动算子，改进整型粒子群优化算法，实现从特征簇中选出代表性特征构成最终的特征子集。在19个非平衡数据集的实验结果表明所设计的算法有效地提高了非平衡数据的分类性能。

关键词: 自适应邻域, 混合采样, 模糊邻域互信息, 特征聚类, 特征选择

Abstract: To solve the problems that the traditional neighborhood rough sets do not consider the class-distribution of imbalanced data, and it is difficult for most neighborhood systems to find the optimal neighborhood radius through manual debugging and the number of clusters needs to be specified in clustering, a feature selection method for imbalanced data based on adaptive neighborhood and clustering is proposed. Firstly, the adaptive K-nearest neighbors and shared nearest neighbors of samples are determined according to the average distance between the samples and other samples under each feature, and then the hybrid sampling model is designed based on adaptive neighborhood density to develop the balanced decision systems. Secondly, a new neighborhood radius is defined based on the feature distribution, the Gaussian kernel function is used to research the fuzzy similarity relationship between samples in the neighborhood. The fuzzy neighborhood mutual information is proposed to measure the correlation between features, and features are clustered based on this. Finally, the particle swarm initialization strategy is designed based on fuzzy neighborhood mutual information. To improve the integer particle swarm optimization algorithm, the dynamic bit mask strategy and the differential perturbation operator suitable for integer coding are introduced, and the representative features are selected from the feature cluster to form the final feature subset. The experimental results on 19 imbalanced datasets show that the developed algorithm can effectively improve the classification effect of imbalanced data.

Key words: adaptive neighborhood, hybrid sampling, fuzzy neighborhood mutual information, feature clustering, feature selection

孙林, 梁娜, 王欣雅. 基于自适应邻域与聚类的非平衡数据特征选择[J]. 计算机工程与应用, 2024, 60(14): 74-85.

SUN Lin, LIANG Na, WANG Xinya. Feature Selection Using Adaptive Neighborhood and Clustering for Imbalanced Data[J]. Computer Engineering and Applications, 2024, 60(14): 74-85.

参考文献

[1] SUN L, LI M M, DING W P, et al. AFNFS: adaptive fuzzy neighborhood-based feature selection with adaptive synthetic over-sampling for imbalanced data[J]. Information Sciences, 2022, 612: 724-744.
[2] 严远亭, 马迎澳, 任艳平, 等. 基于构造性神经网络与全局密度信息的不平衡数据欠采样方法[J]. 计算机科学, 2023, 50(10): 48-58.
YAN Y T, MA Y A, REN Y P, et al. Imbalanded undersampling based on constructive neural network and golbal desity information[J]. Computer Science, 2023, 50(10): 48-58.
[3] 李京泰, 王晓丹. 基于代价敏感激活函数XGBoost的不平衡数据分类方法[J]. 计算机科学, 2022, 49(5): 135-143.
LI J T, WANG X D. XGBoost for imbalanced data based on cost-sensitive activation function[J]. Computer Science, 2022, 49(5): 135-143.
[4] 侯天宝, 王爱银. 基于Stacking特征增强多粒度联级Logistic的个人信用评估[J]. 河南师范大学学报(自然科学版), 2023, 51(3): 111-122.
HOU T B, WANG A Y. Personal credit evaluation based on Stacking feature enhancing multi-grained cascade logistic[J]. Journal of Henan Normal University (Natural Science Edition), 2023, 51(3): 111-122.
[5] SUN L, ZHANG J X, DING W P, et al. Feature reduction for imbalanced data classification using similarity-based feature clustering with adaptive weighted k-nearest neighbors[J].Information Sciences, 2022, 593: 591-613.
[6] 杨洁, 匡俊成, 王国胤, 等. 代价敏感的多粒度邻域粗糙模糊集的近似表示[J]. 计算机科学, 2023, 50(5): 137-145.
YANG J, KUANG J C, WANG G Y, et al. Cost-sensitive multigranulation approximation of neighhorhood rough fuzzy sets[J]. Computer Science, 2023, 50(5): 137-145.
[7] 刘艳, 程璐, 孙林. 基于K-S检验和邻域粗糙集的特征选择方法[J]. 河南师范大学学报(自然科学版), 2019, 47(2): 21-28.
LIU Y, CHENG L, SUN L. Feature selection method based on K-S test and neighborhood rough set[J]. Journal of Henan Normal University (Natural Science Edition), 2019, 47(2): 21-28.
[8] 孙林, 徐枫, 李硕, 等. 基于ReliefF和最大相关最小冗余的多标记特征选择[J]. 河南师范大学学报(自然科学版), 2023, 51(6): 21-29.
SUN L, XU F, LI S, et al. Multilabel feature selection algorithm using ReliefF and mRMR[J]. Journal of Henan Normal University (Natural Science Edition), 2023, 51(6): 21-29.
[9] 陈盼盼, 林梦雷, 刘景华, 等. 基于邻域粗糙集的多标记属性约简算法[J]. 闽南师范大学学报(自然科学版), 2018, 31(4): 1-11.
CHEN P P, LIN M L, LIU J H, et al. Multi-label attribute reduction algorithm based on neighborhood rough set[J]. Journal of Minnan Normal University (Natural Science), 2018, 31(4): 1-11.
[10] SUN L, YIN T Y, DING W P, et al. Feature selection with missing labels using multilabel fuzzy neighborhood rough sets and maximum relevance minimum redundancy[J]. IEEE Transactions on Fuzzy Systems, 2021, 30(5): 1197-1211.
[11] 李顺勇, 王改变, 余曼. 基于相似性特征聚类的加权无监督特征选择算法[J]. 贵州师范大学学报(自然科学版), 2021, 39(1): 49-57.
LI S Y, WANG G B, YU M. Weighted unsupervised feature selection algorithm based on similarity feature clustering[J]. Journal of Guizhou Normal University (Natural Sciences), 2021, 39(1): 49-57.
[12] CAPO M, PEREZ A, LOZANO J A. A cheap feature selection approach for the k-means algorithm[J]. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32(5): 2195-2208.
[13] 王琛, 董永权. 基于二进制灰狼优化的特征选择及文本聚类[J]. 计算机工程与设计, 2021, 42(9): 2526-2535.
WANG C, DONG Y Q. Feature selection based on binary grey wolf optimization and text clustering[J]. Computer Engineering and Design, 2021, 42(9): 2526-2535.
[14] CHATTERJEE I, GHOSH M, SINGH P K, et al. A clustering-based feature selection framework for handwritten indic script classification[J].Expert Systems, 2019, 36(6): 12459.
[15] ZHANG Y, WANG Y H, GONG D W, et al. Clustering-guided particle swarm feature selection algorithm for high-dimensional imbalanced data with missing values[J].IEEE Transactions on Evolutionary Computation, 2022, 26(4): 616-630.
[16] SUN L,WANG L Y,DING W P, et al.Feature selection using fuzzy neighborhood entropy-based uncertainty measures for fuzzy neighborhood multigranulation rough sets[J]. IEEE Transactions on Fuzzy Systems, 2021, 29(1): 19-33.
[17] KENNEDY J.Bare bones particle swarm[C]//Proceedings of the IEEE Swarm Intelligence Symposium, 2003: 80-87.
[18] 李冰晓, 万睿之, 朱永杰, 等. 基于种群分区的多策略综合粒子群优化算法[J].河南师范大学学报 (自然科学版), 2022, 50(3): 85-94.
LI B X, WAN R Z, ZHU Y J, et al. Multi-strategy comprehensive article swarm optimization algorithm based on population partition[J].Journal of Henan Normal University (Natural Science Edition), 2022, 50(3): 85-94.
[19] SUN L, QIN X Y, DING W P. Nearest neighbors-based adaptive density peaks clustering with optimized allocation strategy[J]. Neurocomputing, 2022, 473: 159-181.
[20] 谢娟英, 丁丽娟. 完全自适应的谱聚类算法[J]. 电子学报, 2019, 47(5): 1000-1008.
XIE J Y, DING L J. The true self-adaptive spectral clustering algorithms[J].Acta Electronica Sinica, 2019, 47(5): 1000-1008.
[21] 刘晓金, 陈文武, 王庆锋. 基于优化核函数带宽SVDD的机械振动预警模型[J]. 机电工程, 2023, 40（11）:1641-1654.
LIU X J, CHEN W W, WANG Q F. Mechanical vibration warning model based on optimized kernel bandwidth SVDD[J]. Journal of Mechanical & Electrical Engineering, 2023,40(11):1641-1654.
[22] SONG X F, ZHANG Y, GONG D W, et al. A fast hybrid feature selection based on correlation?guided clustering and particle swarm optimization for high?dimensional data[J]. IEEE Trans Cybernetics, 2022, 52(9): 9573-9586.
[23] LI A D, XUE B, ZHANG M J. Improved binary particle swarm optimization for feature selection with new initialization and search space reduction strategies[J].Applied Soft Computing,2021,106: 107302.
[24] CHEN H M, LI T R, FAN X, et al. Feature selection for imbalanced data based on neighborhood rough sets[J]. Information Sciences, 2019, 483: 1-20.
[25] YU K,WU X D,DING W. Scalable and accurate online feature selection for big data[J].ACM Transactions on Knowledge Discover from Data, 2016, 11(2): 1-39.
[26] ZHOU P, HU X G, LI P P, et al. Online streaming feature selection using adapted neighborhood rough set[J]. Information Sciences, 2019, 481: 258-279.
[27] ZHOU P, HU X G, LI P P, et al. OFS-density: a novel online streaming feature selection method[J]. Pattern Recognition, 2018, 86(2): 48-61.
[28] ZHOU J,FOSTER D P, STINE R A, et al. Streamwise feature selection[J]. The Journal of Machine Learning Research, 2006, 7(9): 1861-1885.
[29] HU Q H,YU D R, LIU J F, et al. Neighborhood rough set based heterogeneous feature subset selection[J]. Information Sciences, 2008, 178(18): 3577-3594.
[30] LIU J F, HUAI Q H, YU D R. A weighted rough set based method developed for class imbalance learning[J].Information Sciences, 2008, 178(4): 1235-1256.
[31] FRANK E,HALL M A,IAN H.The weka workbench[M]. The data mining and knowledge discovery handbook. Cham: Springer, 2016.
[32] MOAYEDIKIA A, ONG K L, BOO Y L, et al. Feature selection for high dimensional imbalanced class data using harmony search[J]. Engineering Applications of Artificial Intelligence, 2017, 57: 38-49.
[33] WU X D,YU K,DING W,et al.Online feature selection with streaming features[J]. IEEE Transcations on Pattern Analysis Machine Intelligence, 2013, 35(5): 1178-1192.
[34] SUN L,ZHANG J X,DING W P,et al. Mixed measure-based feature selection using the fisher score and neighborhood rough sets[J]. Applied Intelligence, 2022, 52(15): 17264-17288.