#1

Member
Santa Rosa - CA
(This post was last modified: 10-25-2019, 05:31 PM by GroomingDept.)
Even if you don't know anything about computing or Artificial Intelligence (AI) or fragrance design, this is some super cool stuff!

[font=-webkit-standard]https://ai.googleblog.com/2019/10/learning-to-smell-using-deep-learning.html[/font]

Matsilainen likes this post
#2
(This post was last modified: 10-26-2019, 03:42 AM by trashcanmagic.)
look google is the best and I personally owe my current career to their contributions to the field via deepmind's publications and tensorflow but this:

Precision 0.379 [0.351, 0.398]
Recall 0.390 [0.365, 0.412]
F1 0.360 [0.337, 0.372]

ok I thought about it some more and it's a multi-class problem with a LOT of classes but even based on their own descriptions of how different similar graph structures could be in their smells it still strikes me as far from usable outside of a research project.


Users browsing this thread: