Can vary quite a lot but generally answered by how quickly you can identify the right data to use for training and then if you can get enough of it (seen this take weeks, also seen it take 6 months)
@ coding peeps How difficult is it to code & train a ML algorithm to do a task with reasonable accuracy? How does this time/difficulty differ between tasks such as those that require NLP? Just realised I have no concept as to the entry barrier or resources needed lol
-
-
-
Mm I thought that it would've been very data centric In that case, it could be that better/more prolific data capture would be better for advancing AI than AI research itself :p
- Näytä vastaukset
Uusi keskustelu -
-
-
It would take minimum 4 months of coding that particular language every day to get to a point like that. But it wouldn't be particularly difficult.
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
A lot of it depends on the training data. The actual code can be very short (but complex). Training time varies a lot too. SVMs are pretty quick to train. Big neural nets take a long time to train. But there are services online that let you upload CSVs and auto-train like bigML
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
-
-
NLP takes a bit for some people to wrap their head around. Image based stuff is a lot more intuitive because you can visualize the learned features. There's challenge is knowing how to encode your data well, and get good training data that represents the task at hand.
Kiitos. Käytämme tätä aikajanasi parantamiseen. KumoaKumoa
-
Lataaminen näyttää kestävän hetken.
Twitter saattaa olla ruuhkautunut tai ongelma on muuten hetkellinen. Yritä uudelleen tai käy Twitterin tilasivulla saadaksesi lisätietoja.