Loading video...

Video Failed to Load

Go Home

32 Year-old Yann LeCun shows off the world's first Convolutional Network. This is 1993.

408,466 views • 2 years ago •via X (Twitter)

11 Comments

Yann LeCun's profile picture
Yann LeCun2 years ago

Akshually, this is from 1989, and I'm 29.

ₕₐₘₚₜₒₙ — e/acc's profile picture
ₕₐₘₚₜₒₙ — e/acc2 years ago

You were 29 when you pressed the button, and you were 32 years old when you got the results back😹🤝

Krzysztof Woś's profile picture
Krzysztof Woś2 years ago

@ylecun Not quite the first. TDNN was there in the 1980s.

Alex Honchar's profile picture
Alex Honchar2 years ago

@ylecun Technically in 1980 we had Fukushima’s “neocognitron” which we can consider as a CNN as well, @ylecun maybe you could correct if this was actually an inspiration for convolutional neural nets?

ₕₐₘₚₜₒₙ — e/acc's profile picture
ₕₐₘₚₜₒₙ — e/acc2 years ago

So cool

Marina🌰🌱🌿🎋's profile picture
Marina🌰🌱🌿🎋2 years ago

@codexeditor @ylecun Now we know where the “Hello World” equivalent in ML came from.

ₕₐₘₚₜₒₙ — e/acc's profile picture
ₕₐₘₚₜₒₙ — e/acc2 years ago

So real

Michael Tomlinson ✨'s profile picture
Michael Tomlinson ✨2 years ago

Meh? I did something similar using plain neural nets and a B&W CMOS camera in 1986 when still in high school. 16-bit FP, no bias nodes, no hidden layers… learned the alphabet in about 40 training cycles. All based on a 1-paragraph description of ‘weights and nodes’ from a Scientific American article I found. Neural nets are neat.

Prashant's profile picture
Prashant2 years ago

@ylecun Did he know back then he would be having a social media love story with Elon Musk? 😂

Alsa ⚡️'s profile picture
Alsa ⚡️2 years ago

@ylecun @ylecun he was my biggest inspiration for machine learning ❤️

ₕₐₘₚₜₒₙ — e/acc's profile picture
ₕₐₘₚₜₒₙ — e/acc2 years ago

Same

Related Videos