I really, really, really want to create a machine learning & type theory workshop, but I think the communities look at one another with suspicion and condescension. But I think both communities actually need each other so that's bad
Always been a bit intimidated by machine learning, but yeah, having some kind of 'type theoretic machine learning' might make me less scared and intimidated!
I've also had success talking to machine learning/data scientist people about types by explaining how type systems could help with 'expensive data pipeline breaking 12 hours in on a silly typo'.
But I understand you're probably thinking more about how types can be applied to the models themselves in order to improve results of training. Or something like that? 😅