欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
一单元复数参考独立成分分析算法存在阈值参数难以确定的问题。通过将算法的目标优化函数巧妙地调整为期望提取信号的幅值和参考信号的近似性量度，基于机器学习原理和经典的Kuhn-Tucker条件提出一种改进的固定点算法，有效避免人为选取选择阈值参数和步长参数，降低了计算复杂度，并提高了算法收敛的稳定性和收敛速率。针对复数合成数据的仿真实验证实了所提算法的有效性。
英文摘要：
It is difficult to determine the threshold parameter of the one-unit complex-valued independent component analysis with refer- ence（ICA-R） algorithm. An improved fixed-point algorithm based on machine learning and classical Kuhn-Tucker condition is proposed by taking the closeness measure of the magnitude of the desired signal and reference singnal as the contrast function, it can avoid selecting the threshold parameter and the step parameter and reduce the computional complexity to accelerate the convergence. Computer simulations with complex--valued synthetic signals confirm its validity and stability.
同期刊论文项目
同项目期刊论文
Performance analysis of LSEs for 2-D harmonic signals in zero-mean multiplicative and additive noise
Performance analysis of least squares estimation for harmonic signals in zero-mean multiplicative an
A statistically efficient algorithm for estimating the frequencies of 2-D Harmonics in multiplicativ
An efficient and fast algorithm for estimating the frequencies of 2-D superimposed exponential signa
MTSI algorithm for frequencies estimation of 2-D superimposed exponential model in multiplicative an
Noise subspace approach for frequency estimation of two-dimensional harmonics in zero-mean multiplic
Hyperparameters optimization for support vector machine with generalized Gaussian kernel using repea
Performance analysis of least squares estimation for harmonic signals in multiplicative and additive
期刊信息
位置：