摘要：
In the traditional data fragment recognition algorithm,the influence of the fragment＇s own attribute to the algorithm is often ignored,which leads to the low accuracy of the data fragment recognition.Based on the combination of LRFU strategy and association analysis method,it proposes a large database fragment classifica-tion and recognition algorithm for unbalanced data.Based on resampling algorithm,it illustrates the unbalance data fragments,adds the filter coefficients to the middle of the filter coefficients according to the sampling re-sults,calculates the convolution of the unbalanced data fragments in the filter banks,rebuilds the results of the convolution calculation,and obtains the fragment characteristic sequence.Using the similar piecewise linear method to deal with the unbalanced data fragments with high convergence degree in the sequence,it realizes the fragments.The conversion of objective function combined with the expansion matrix obtains the fitness function.Using LRFU strategy,it schedules the fitness function,combines the fusion association analysis method to deter-mine the unbalance data fragment attribute value,completes the recognition of unbalanced data fragments.The experimental results show that the fitness and recognition accuracy of the improved method are higher than that of the others,and it has some advantages.%传统数据碎片识别算法中,往往会忽略碎片自有属性对算法的影响,导致数据碎片识别准确率较低,为此提出了基于LRFU策略与关联分析方法相结合的大数据库不均衡数据碎片分类识别算法.通过重采样算法进行不均衡数据碎片升降采样;依据采样结果对滤波器系数中间补零,并用滤波器组对不均衡数据碎片进行卷积计算,通过对卷积计算结果进行重构处理,获取碎片特征序列.采用类似分段线性法处理序列中汇聚程度较高的不均衡数据碎片,实现碎片分类,通过目标函数转换,并结合扩张矩阵得到适应度函数;采用LRFU策略对适应度函数进行调度,融合关联分析方法,确定不均衡数据碎片属性值,实现不均衡数据碎片的识别.实验结果表明,采用改进方法进行不均衡数据碎片分类识别时,其适应度值与识别准确率较高,具有一定的优势.
