Note: The context here is supervised regression
Without having to change it manually in the source code, is the uncertain_input a feature that can be turned off manually by the user? Or rather is it a feature that's intentionally hidden from the user?
Also, with regards to the inducing variables Z, how would I set it for the first ObservedLayer and fix it such that the subsequent layers would then follow suit and update itself w.r.t to the first layer?