I was just going to post that you could probably make an autoencoder over flowers and use it as an embedding to probe genomic alterations that tweak the flower generator functions
Nope. DNA is not really useful for show boundaries between species, because we all share a lot of it and an only a very specialist expert would know what genes make the flower. Is also very expensive and time consuming.
Plant families have consistent, easily recognizable and often stable floral formulas [1], that don't change among entire tribes. Finding an interval of families to classify our sample is much easier this way, does not need expensive machines and can be done in the middle of a jungle at 1000 Km of the next laboratory. In any case floral formulas are just a small part in the whole picture.
[1] This was the master plan; and now here come the tropics with its own agenda.
Taking botany was up there with physics and chemistry as one of the best and most impactful courses I've taken.
It feels like a real phenomenon, there's knowledge that can be learned, and certainly has a Eureka moment once you start seeing it.
Social sciences and mathematics feel more about learning stuff other people said, and the latter more about preparing to do useful stuff. The natural science just have that undeniable factual knowledge that is transferred and my brain just accepts never to let it go, as it doesn't come from humans, it's not political, it's truth.
And it's not one of the boring subjects like geology or edafology, there's an undeniable psychic problem with talking about, sitting in a lecture listening about, looking at and thinking about rocks that plants just don't have. Ok, sorry minerals, I still don't care.
To me plants where just always green stuff, I doubt I could have even surmised a fraction of the knowledge just by observing them carefully even if I tried.
The Voynich manuscript is peculiar, because the plants depicted are familiar but clearly fake to the eyes of every botanist. To describe an imaginary a plant is easy. To describe a convincing imaginary plant is much harder and Voynich failed on that.
AI has exactly the same problem, when the program draws people without really understanding first what is a human. Seven fingers? not from this planet.
It would be wild if flowers could be expressed in a numerical system and floral formulas / operations could form a ring or a group in the mathematical sense!
Floral diagrams are essentials IMO along with floral formulas to read a flower.
No clue what I'm doing but the flower diagrams it generates from floral formula parameters are pretty
https://kvetnidiagram.8u.cz/index_en.php