Deep Residual Learning for Image Recognition, image 1 of 11 / 1
Item

Deep Residual Learning for Image Recognition

in Papers I Reread by Lena Carver

Deep Residual Learning for Image Recognition on “Papers I Reread”, a list by Lena Carver on TheLysts.

Details

Photo
—
Name
—
Field
Machine learning / computer vision
Year
2015
Lead author
Kaiming He
Core idea
Introduces residual networks (ResNets), using skip connections to train very deep neural nets effectively.
Why it hooked me
The trick is conceptually simple but unlocked a whole new performance tier—my favorite kind of hack.
Best for
Getting a feel for how modern deep nets are actually architected, not just buzzworded.
Length feel
Conference‑paper length; figures and results tables do heavy lifting.
Vibe
Efficient, empirical, no drama—just “here’s a clever idea and the leaderboard receipts.”
Nerdy flex
Referencing ResNets when someone says “AI is basically magic” is deeply therapeutic.