AI trained on old scientific papers makes discoveries humans missed
6 months, 1 week ago by kyle | 5 facts | 1 comments | 0 votes

  • Using just the language in millions of old scientific papers, a machine learning algorithm was able to make completely new scientific discoveries. -
  • “It can read any paper on material science, so can make connections that no scientists could,” researcher Anubhav Jain said. -
  • The algorithm used was word2vec -
  • Using words found in scientific abstracts, the algorithm was able to understand concepts such as the periodic table and the chemical structure of molecules. -
  • To train the algorithm, the researchers assessed the language in 3.3 million abstracts related to material science, ending up with a vocabulary of about 500,000 -

kyle - 6 months, 1 week ago
To be honest it seems like they just trained word2vec on a bunch of papers and are saying that it found some useful relationships. I fail what is so groundbreaking here. People have some similar things with Wikipedia and CommonCrawl data.

Story Info
  • Upvotes: 1
  • Downvotes: 1
  • Poster: kyle