There are many contexts in which we'd like to model some underlying process that doesn't necessarily need autoregressive dynamics. Shared latent states for one, or accommodating sampling uncertainty for another. Or perhaps we want to use a GP of time without any extra AR parameters
There are many contexts in which we'd like to model some underlying process that doesn't necessarily need autoregressive dynamics. Shared latent states for one, or accommodating sampling uncertainty for another. Or perhaps we want to use a GP of time without any extra AR parameters