types of scientific paper
- we applied machine learning to this existing problem and it worked kind of well, like, not well enough to replace the existing state of the art, but kind of impressive, right, we didn't give it any domain-specific knowledge or anything