Tessera
S
Folgen

Stephane Deny

@stphtphsn.bsky.social

Assistant Professor at Aalto. Neuro - ML https://scholar.google.com/citations?hl=fr&user=Z3jU6mQAAAAJ&view_op=list_works&authuser=1&sortby=pubdate

39 Posts346 Follower774 Folgt
S
Stephane Deny@stphtphsn.bsky.social · 9mo ago

New preprint! "A Deep Learning Model of Mental Rotation Informed by Interactive VR Experiments" 🔗 arxiv.org/abs/2512.13517 by Raymond Khazoum, Daniela Fernandes, Aleksandr Krylov, Qin Li and myself

Bild
Ray@rajky.bsky.social · 9mo ago

A striking example of spatial reasoning is humans’ ability to recognize objects across viewpoints—a skill known as mental rotation. How does the brain achieve this ability? To answer this, we propose a mechanistic model of human mental rotation. Paper: arxiv.org/abs/2512.13517 1/🧵

Jonathan Kadmon@kadmonj.bsky.social · last yr.

(1/6) Excited to share a new preprint from our lab! Can large, deep nonlinear neural networks trained with indirect, low-dimensional error signals compete with full-fledged backpropagation? Tl;dr: Yes! arxiv.org/abs/2502.20580.

Training Large Neural Networks With Low-Dimensional Error Feedback

Training deep neural networks typically relies on backpropagating high dimensional error signals a computationally intensive process with little evidence supporting its implementation in the brain. Ho...

arxiv.org

S
Stephane Deny@stphtphsn.bsky.social · 2y ago

New Preprint: "On the Ability of Deep Networks to Learn Symmetries from Data: A Neural Kernel Theory" by Andrea Perin and myself. Retweets of this post or the thread below are highly appreciated! Follow @zazzarazzaz.bsky.social for future updates on this line of work.

Bild
Andrea Perin@zazzarazzaz.bsky.social · 2y ago

Little is known about how deep networks interact with structure in data. An important aspect of this structure is symmetry (e.g., pose transformations). Here, we (w/ @stphtphsn.bsky.social) study the generalization ability of deep networks on symmetric datasets: arxiv.org/abs/2412.11521 (1/n)

Thùy Vy T Nguyễn@thuyvytnguyen.bsky.social · 2y ago

I did my PhD with Ed Deci, who usually says very few words. In our meetings, it was his reserved nature that gradually allowed me to come out of my shell, be my own person. That experience teaches me one valuable lesson now that I mentor others: know when to shut up.

S
Stephane Deny@stphtphsn.bsky.social · 2y ago

New Preprint: "On the Ability of Deep Networks to Learn Symmetries from Data: A Neural Kernel Theory" by Andrea Perin and myself. Retweets of this post or the thread below are highly appreciated! Follow @zazzarazzaz.bsky.social for future updates on this line of work.

Bild
Andrea Perin@zazzarazzaz.bsky.social · 2y ago

Little is known about how deep networks interact with structure in data. An important aspect of this structure is symmetry (e.g., pose transformations). Here, we (w/ @stphtphsn.bsky.social) study the generalization ability of deep networks on symmetric datasets: arxiv.org/abs/2412.11521 (1/n)

Andrea Perin@zazzarazzaz.bsky.social · 2y ago

Little is known about how deep networks interact with structure in data. An important aspect of this structure is symmetry (e.g., pose transformations). Here, we (w/ @stphtphsn.bsky.social) study the generalization ability of deep networks on symmetric datasets: arxiv.org/abs/2412.11521 (1/n)

On the Ability of Deep Networks to Learn Symmetries from Data: A Neural Kernel Theory

Symmetries (transformations by group actions) are present in many datasets, and leveraging them holds significant promise for improving predictions in machine learning. In this work, we aim to underst...

arxiv.org

Guillaume Bellec@bellecguill.bsky.social · 2y ago

Pre-print 🧠🧪 Is mechanism modeling dead in the AI era? ML models trained to predict neural activity fail to generalize to unseen opto perturbations. But mechanism modeling can solve that. We say "perturbation testing" is the right way to evaluate mechanisms in data-constrained models 1/8

Bild