【摘要】 以提升不平衡数据集分类检测为研究目标，提出基于改进级联算法的不平衡数据集分类检测算法.首先，采用卡尔曼滤波法进行数据去噪预处理，利用小波阈值去噪算法二次消除噪声数据，并对去噪结果进行归一化预处理；利用DPC算法提取数据的局部密度特征，利用时间编码挖掘数据的时序性特征，采用Apriori算法的强关联规则提取数据集特征；利用模糊层次聚类算法对支持向量机进行优化，实现数据类型的划分；利用改进的级联算法联合布谷鸟算法实现不平衡数据集分类检测.实验结果表明本方法的分类协方差低于0.15，检测准确率高于95%，检测时间低于2.2 ms，有效提升了不平衡数据集分类检测效果.更多还原

【Abstract】 With the research goal of improving the classification and detection of imbalanced datasets, a classification and detection algorithm for imbalanced datasets based on an improved cascade algorithm is proposed. Firstly, the Kalman filtering method is used for data denoising preprocessing, and the wavelet threshold denoising algorithm is used to eliminate noisy data twice, and the denoising results are normalized for preprocessing; extracting local density features of data using DPC algorithm, mining temporal features of data using time encoding, and extracting dataset features using strong association rules of Apriori algorithm; using fuzzy hierarchical clustering algorithm to optimize support vector machines and achieve data type partitioning; utilizing an improved cascaded algorithm combined with the cuckoo algorithm to achieve imbalanced dataset classification detection. The experimental results show that the classification covariance of this method is less than 0.15, the detection accuracy is higher than 95%, and the detection time is less than 2.2 ms, effectively improving the classification and detection performance of imbalanced datasets.更多还原