Inverse Problems with Closed Form Expression

1

Hello,
I was wondering if there are any other popular inverse models (that are not in brainstorm) which utilize closed form expression to compute the inverse model?

I understand that this means the model needs to be $L_2$ based, but I was wondering if there is any paper(s) working on prior assumptions we could add to $L_2$ based solutions get better results?

Thanks

2

hi,

If you want a linear inverse you'll need indeed to have an squared L2 regularization.
Now you can use a weighted L2 norm and "learn" the weights. That's pretty much
what hierarchical bayesian model with full MAP estimation does or what the sparse
bayesian learning with type-II maximum likelihood ends up doing to.

Hope this helps,
Alex

1 Like