News for hackers. Stuff that matters.

A comment on Homomorphically encrypted CIFAR-10 inference in 200ms. Parent comment

Re:Homomorphically encrypted CIFAR-10 inference in 200ms by sambejk on Friday July 17, 2026 @03:03PM (#48951062)
I think calculator is not part of the labels in cifar-10. Pretty normal as it has ‘only’ 90% accuracy. Things could improve quickly though
[ Parent | on HN ]
Re:Homomorphically encrypted CIFAR-10 inference in 200ms by jcs on Friday July 17, 2026 @03:42PM (#48951478)
That 90% only measures images already known to be one of CIFAR-10’s ten classes. There is no “none of the above,” so even a perfect benchmark score would still force a calculator into one of those labels.
[ Parent | on HN ]
Re:Homomorphically encrypted CIFAR-10 inference in 200ms by Eduard on Friday July 17, 2026 @04:26PM (#48951917)
yep: https://en.wikipedia.org/wiki/CIFAR-10

"The 10 different classes represent airplanes, cars, birds, cats, deer, dogs, frogs, horses, ships, and trucks."

[ Parent | on HN ]