2338 shaares
Perspective intéressante. Se résume en quelques phrases:
* Companies brag about the size of their datasets the way fishermen brag about the size of their fish
* But even big companies only use a tiny fraction of the data they collect.
* Typical deep-learning models only work on massive amounts of labeled data. And labelling a large dataset takes hundreds of thousands of dollars and months of time. (...) Too many smaller companies don’t realize this and acquire massive data stores that they can’t afford to use.
* big data isn’t big, but good data is even smaller
En résumé, il y a effectivement des enjeux sur le big data. Mais principalement dans les entreprises qui investissent énormément dans le big data (google, facebook, etc.). Pour la majorité, on n'en est pas encore là.
* Companies brag about the size of their datasets the way fishermen brag about the size of their fish
* But even big companies only use a tiny fraction of the data they collect.
* Typical deep-learning models only work on massive amounts of labeled data. And labelling a large dataset takes hundreds of thousands of dollars and months of time. (...) Too many smaller companies don’t realize this and acquire massive data stores that they can’t afford to use.
* big data isn’t big, but good data is even smaller
En résumé, il y a effectivement des enjeux sur le big data. Mais principalement dans les entreprises qui investissent énormément dans le big data (google, facebook, etc.). Pour la majorité, on n'en est pas encore là.