双变换算法在多维序列数据分析中的优化研究

Research on optimization of double transform algorithm  in multidimensional sequence data analysis

摘要:

在流数据中,降低维度是处理多维序列数据的重要因素.提出一种双变换算法（DTA）,针对在线序列数据,分别进行酉变换和双曲线旋转变换的双变换处理,得到假设函数的参数,通过牛顿算法迭代预测误差值,直到小于所预设的阈值,从而得到最优预测值.仿真结果表明,对比OGD和RON两种算法,DTA算法在保障算法稳定性的前提下,有效减少计算时间.

Abstract:

In streaming data, dimension reduction is an important factor in processing multidimensional sequence data. In this paper, a double transform algorithm (DTA) is proposed. For the online sequence data, unitary transformation and hyperbolic rotation transformation is carried out respectively in DTA，and the parameters of the hypothesis function are obtained. The error values are predicted by the Newton iterative algorithm, and the optimal prediction value is obtained until the error is less than the predefined threshold. The simulation results show that, compared with the two algorithms of OGD and RON, the DTA algorithm effectively reduces the computation time under the premise of ensuring the algorithm stability.