Тёмный

DiffTaichi: Differentiable Programming for Physical Simulation 

Ethan (Yuanming) Hu
Подписаться 2,1 тыс.
Просмотров 17 тыс.
50% 1

Опубликовано:

 

22 окт 2024

Поделиться:

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист
Посмотреть позже
Комментарии : 9   
@davisburnside9609
@davisburnside9609 4 года назад
Came here from Two Minute Papers. This is insane & you are brilliant!
@xuandayang7116
@xuandayang7116 4 года назад
The water renderer case is really cool!
@MrNoipe
@MrNoipe 4 года назад
Awesome work, its great you made reproducing your work simple to do.
@toafloast1883
@toafloast1883 5 лет назад
time to pick up taichi
@SamA-kb9ug
@SamA-kb9ug 4 года назад
The legend himself!
@neurophilosophers994
@neurophilosophers994 4 года назад
How did it trick the AI into thinking the squirrel was a goldfish ? It had to optimize toward a goal of matching data from that which the AI had already classified as goldfish worthy ? Or ? Confusing
@technokicksyourass
@technokicksyourass 4 года назад
I would guess, by computing a gradient in the goldfish direction, then backpropagating, then updating parameters on the water ripples, to create distortions that push the classifier in the goldfish direction. If that makes sense...
@welcomeaioverlords
@welcomeaioverlords 4 года назад
See: "adversarial examples"
@anthonyyao5889
@anthonyyao5889 4 года назад
跑了历程,挺走趣的,正要尝试3d
Далее
Handsoms😍💕
00:15
Просмотров 6 млн
Differentiable Programming with Julia by Mike Innes
23:56
[SIGGRAPH Asia 2019] The Taichi Programming Language
7:55
Why Does Diffusion Work Better than Auto-Regression?
20:18
Compiled Python is FAST
12:57
Просмотров 113 тыс.
Finally, Differentiable Physics is Here!
5:25
Просмотров 365 тыс.