iiinova — where your digital self keeps evolving

Antoine Deleforge

@adeleforge.bsky.social

Research scientist at #Inria. Audio signal processing, Acoustics, Machine Learning, Bicycle Riding, Lindy Hop Dancing.

Antoine Deleforge @adeleforge.bsky.social · 30/04/2026

Shooting an interview for a future permanent exhibition at the "Palais de la découverte" museum of Paris, on the theme "Can One Hear the Shape of an Object"?...wait and see 👀 Also, who said researcher life was dull? 😅 I am surrounded by ~150k€ of audiovisual equipments, better not trip on a wire! 😬

000

Antoine Deleforge @adeleforge.bsky.social · 16/02/2025

[The 3D plot of ReLU above is from addxorrol.blogspot.com/2024/07/some... ]. Guido Montufar has an excellent paper showing that the number of regions in the piecewise-affine function defined this way is exponential in the depth of the network: proceedings.neurips.cc/paper_files/... 3/3.

addxorrol.blogspot.com

000

Antoine Deleforge @adeleforge.bsky.social · 16/02/2025

I will use that for the deep learning class I teach. Feel free to reuse ! In general I love the "origami effect" explanation of why deep learning works well. This is particularly clear for "ReLU" non linearities. You can then view each layer as folding your space in two. 2/n

100

Antoine Deleforge @adeleforge.bsky.social · 16/02/2025

I just had a lot of fun creating this animation, inspired by a quote from @fchollet.bsky.social 's book "Deep Learning with Python" 😊. You can view the layers of a DNN as progressively unfolding the intertwined manifolds of the classes you want to separate, like two crumbled sheets of paper. 🧵 1/n

120

Antoine Deleforge @adeleforge.bsky.social · 20/11/2024

Hi Jonathan! 🙂

110

Antoine Deleforge @adeleforge.bsky.social · 20/11/2024

Hello there! Checking out whether the sky is bluer here 🙂

130