Systems Smart Enough to Know When They're Not Smart Enough

Re:Systems Smart Enough to Know When They're Not Smart Enough by jtraffic on Tuesday March 14, 2017 @02:23PM (#13870069)

I skimmed. My synopsis: "ML models should report confidence levels."

[ Parent | on HN ]

Re:Systems Smart Enough to Know When They're Not Smart Enough by pimmen on Wednesday March 15, 2017 @03:20AM (#13874387)

Or, like Watson when it played Jeopardy, not respond at all if the confidence is under a certain threshold.

[ Parent | on HN ]