5 ms·
More users -> more data -> better product -> more users s/b More users -> more labeled data -> better product -> more users ML systems need lots of labeled d
by jgalt212 9y ago
More users -> more data -> better product -> more users
s/b
More users -> more labeled data -> better product -> more users
ML systems need lots of labeled data, not just lots of data. This is one of the primary why game playing AI's have had such great successes, tons of labeled data are relatively cheap. Great discourse on this and other related issues here:
https://medium.com/@karpathy/alphago-in-context-c47718cb95a5 https://medium.com/@karpathy/alphago-in-context-c47718cb95a5
- mgummelt 9y agoSupervised learning requires labels. Unsupervised doesn't. Self driving cars use elements of both. This is why companies like Cruise and Tesla are rushing to get basic self driving cars out there. They need driving data.
- jgalt212 9y agoFair enough. I guess I'm in the camp that unsupervised learning is of limited utility.
- jmalicki 9y agoYou can get creative about labels. For instance, with known velocity of the car, predict where an object in view of the camera will be 5 frames in the future... This makes all data recorded by the camera effectively labeled, and is actually a useful thing to have.
- nl 9y agoHuman input in normal cars with sensors provides supervision.