一种改进的ICA算法及其在fMRI信号上的应用

摘要:给出了独立成分分析(ICA)的一个优化模型,在此基础上,提出了一种牛顿型迭代算法,为加快算法的收敛速度,对牛顿迭代进行了进一步修正,使该算法收敛速度达到三阶.本文从理论上阐明了新方法的合理性和优越性,同时将其应用于实际fMRI数据,经与其他两个ICA算法(Fast ICA算法、infomax算法)比较,该算法能够很好地分离出任务成分,同时大大减少了运算量,提高了运算速度,对处理大数据量的fMRI信号有明显的优势.

An improved ICA algorithm and its application on fMRI data

Abstract: In this paper, an optimization model for independent component analysis (ICA) is presented. A new Newton iteration algorithm based on the model is proposed. In order to fasten the convergence, an improvement on Newton method is made which makes the convergence of the new algorithm is cubic. The effectiveness and advantage are elucidated through theoretical analysis. By applying the algorithm and two other algorithms (Fast ICA and infomax) to invivo fMRI data, the results show that the new algorithm separate independent components from fMRI data very well.and it has fastest convergence speed and lest computation than the other two algorithms. The algorithm has obvious advantage in precessing fMRI signal with huge data.