Hacker News new | ask | show | jobs
by pella 3108 days ago
summary(slides): http://dawn.cs.stanford.edu/assets/dawn-overview.pdf
1 comments

The link to Snorkel [1] is really interesting, labeling data in a low quality programmatic way, which is then fed through a neural network to produce high quality labels is really smart.

[1] https://github.com/HazyResearch/snorkel

Yes, Snorkel and DeepDive look extremely useful. At my job we have a lot of data but it's unlabeled, it will cost millions of dollars to outsource it to India for labeling/data entry.