这是一个贝叶斯独立分量分析(ICA)算法的线性瞬时混合高斯噪声模型和添加剂。解决问题的是ML-II推论,即资源的整合在发现源后和噪声协方差矩阵和混合了最大化的边际似然。充分统计量的估计平均场或变分理论和线性响应修正或通过自适应平均场理论水龙头。平均场方程,解决了信仰传播法的或连续的迭代。-This is a bayesian independent component analysis (ICA) algorithm of instantaneous linear mixed gaussian noise model and additives. The solution to the problem of is ML-II inference, namely of the integration of resources and noise source in found that after the covariance matrices and mixed the maximization of marginal likelihood. Sufficient statistics estimate of the mean field or variational theory and linear response correction or through the adaptive average field theory faucet. Average field equations are solved belief propagation method or continuous iteration.