speaker-0: Prior dissertation is just a fancy term to say, okay, we ask a bunch of experts to tell us what are meaningful ranges, let's say, for parameters. And the realization in a lot of large body of this work has been that typically you can't just ask experts, give me a prior distribution. Okay, because that's not how human knowledge is typically represented. It's not represented in a strictly formal probabilistic way. It's represented sometimes in tacit structures. So elicitation is is a translation process, basically. And so the perennial problem there is by translating this tacit knowledge or non probabilistic knowledge into prior distributions that conform to the expectation, which may not be obvious. An expert can even tell you about this, these are ⁓ the priors that you should use, but then these priors when you actually do prior checking or model checking, they lead to surprising results. For example, your model diverges a lot. Okay? So the idea here was to automate this with with the simulation-based procedure in which okay, you you still ask your experts or expert to elicit certain quantities. Let's say the expert gives you quantiles. a mean and that's all you got. So what you can then do is you can parameterize your priors with what we call hyper parameters. For example, instead of fixing ⁓ the mean and the standard deviation of the Gaussian prior, you let the vary and you start simulating from this hyper model. And when you simulate this you also compute the at the model implied elicit quantities. as a constant we we compute the model implied quantas and means, let's say. And we basically compare them to the ones given by the expert, right? Initially they'll be very far apart. And so we need a way now to propagate this error. The the divergence between the model implied and the expert implied quantities. And if you can keep your pipeline differentiable end to end, that's nice. You use gradient descent, right? You can do where it's reparameterization tricks to make sampling differentiable, etc. If you can't keep the whole pipeline differentiable, you can use black box optimization, something like ⁓ baseline optimization. And you run this iterative procedure and you end up with ideally with the hyperparameters of the prior that most closely match what the expert had in mind. So that's that was the idea of the first paper. and we showed that this this can be very useful for a bunch of case studies. Now in the follow up paper, we decided to take this one step further. In the first approach, it's true that you s you have some flexibility, but you're still confined to a certain parametric family. Right? Yeah, you may optimize for the hyperparameters of a Gaussian, but you're still in a Gaussian world. So we asked here, can you do better? Okay. Can you have some sort of a non-parametric procedure that is is going to give you the prior directly without assuming its distribution of family a priori. And to do this, of course, we turn to generative AI. So it would would basically replace the prior with an untrained generative network. We use normalizing flows for the proof of concept and we applied exactly the same procedure now, but with a normalizing flow as a surrogate for the prior. Okay. And this sounds really cool and all, but of course there are some caveats, right? Because this ⁓ this prior is very flexible, there is a lot of ⁓ non-identifiability issues. So you you you you can have cases where you you end up with ⁓ a lot of possible solutions ⁓ and you have to deal with stability issues so but I think that the general idea is still still very viable because like if you ⁓ if you can take this approach now and add some further constraints you end up with with an end to end pipeline again for finding priors based on non-probabilistic expert knowledge which yeah in two thousand twenty six can also be theoretically followed by an agent. So that is the basic idea. Yeah. Yeah this is this is super interesting and practical I think