大数据特征集准确检测仿真研究

Accurate Detection Simulation Data Feature Set

对大数据特征集进行准确检测,可提高大数据的处理和分析能力。进行大数据特特征集检测时,主要以大数据信息的关键特征为基础,对大数据特征集进行进行准确描述完成检测,而传统的基于Relief的检测方法主要是通过对选取的大数据初始群体特征进行随机选择,导致大数据特征集中关键特征提取不准确,降低了特征集检测的准确性。提出采用粒子量化融合算法的大数据特征集检测方法,通过粒子群算法和离散二进制粒子群算法,提取具有代表性的初始大数据,并计算适应度函数及适应度值,对大数据特征集最优值进行提取和更新。采用速度与位置更新公式,对数据特征进行更新,将大数据特征集检测问题转换为判断其是否达到最优解的问题,完成大数据特特征集的准确检测,仿真结果表明,改进的大数据特征集检测方法精度高、耗时短。

To detect its feature set accurately can improve the processing and analyzing ability of big data. The feature set is described accurately to complete detection based on key feature of big data information during detection.However traditional method mainly selects the initial population characteristic randomly based on Relief. It makes the centralizing extraction of key feature inaccurate and reduces the precision. In this paper we proposed a detection method of big data feature set based on particles quantization fusion algorithm. Firstly,we extracted representative initial big data with particle swarm algorithm and discrete binary particle swarm algorithm. Then we calculated its fitness function and fitness value based on the data mentioned above to extract and update the optimal value of big data feature. The data feature was updated using velocity and position to update function. Finally,we converted the detection problem to judging optimal solution problem and achieved accurate detection of big data feature set. The simulation results show that the modified detection method has higher precision and consumes less time.