i've always just seen it as a wrapper for updating multivariate linear models when the inputs have noise that can be modeled with gaussians. linear models mean

i've always just seen it as a wrapper for updating multivariate linear models when the inputs have noise that can be modeled with gaussians. linear models meaning anything that can be described as a matrix vector multiply. (which is a lot) but i'm no expert.