machine learning algorithm designs antibodies to target specific diseases, then automated robotic systems build and grow them in lab, run tests, and feed data back into algorithm, all with limited human supervision.
So it's a brute force approach using automated systems. They mention their method is superior to traditional brute force methods by doing unorthodox things, but the article does not go into detail into how.
I mean, great news if this methodology pans out. There just very little to go off from the article. Either way, seems like a pretty neat testing suite.
It’s one of those situations where the lack of preconceived notions from a ml algorithm allows it to try a bunch of stuff humans would often discard off hand, which sometimes pans out.
It’s one of the best things about ML based systems that they can make mistakes faster than humans, so they can re-validate or correct our existing understanding of the topic.
Add comment