Sebastian Goldt
@sebgoldt.bsky.social
Theory of neural networks at SISSA in Trieste, Italy
Our workshop on creativity & generative AI just got underway this morning ! Organised with our colleagues Elisa Gremmo and Gabriele Tomasi from the philosophy department at the University of Padova, we will discuss generative AI models and whether they are, in some sense, "creative".
July 3, 2025 at 9:38 AM
Our workshop on creativity & generative AI just got underway this morning ! Organised with our colleagues Elisa Gremmo and Gabriele Tomasi from the philosophy department at the University of Padova, we will discuss generative AI models and whether they are, in some sense, "creative".
Really happy to see this collab with Nishil Patel, Seb Lee, @stefsm.bsky.social and @saxelab.bsky.social finally out!
We apply the statistical physics toolbox to analyse a simple model of reinforcement learning, and find some cool effects, like a speed-accuracy trade-off for generalisation 🚀
We apply the statistical physics toolbox to analyse a simple model of reinforcement learning, and find some cool effects, like a speed-accuracy trade-off for generalisation 🚀
Our paper just came out in PRX!
Congrats to Nishil Patel and the rest of the team*
TL;DR: We analyse policy learning through the lens of statphys, revealing distinct scaling regimes with sharp transitions.
🔗 journals.aps.org/prx/abstract...
*Seb Lee @sebgoldt.bsky.social @saxelab.bsky.social
Congrats to Nishil Patel and the rest of the team*
TL;DR: We analyse policy learning through the lens of statphys, revealing distinct scaling regimes with sharp transitions.
🔗 journals.aps.org/prx/abstract...
*Seb Lee @sebgoldt.bsky.social @saxelab.bsky.social
RL Perceptron: Generalization Dynamics of Policy Learning in High Dimensions
A solvable model for reinforcement learning, the RL perceptron, provides a mathematical framework to analyze learning dynamics, revealing key efficiency factors and a speed-accuracy trade-off that can...
journals.aps.org
May 16, 2025 at 6:13 AM
Really happy to see this collab with Nishil Patel, Seb Lee, @stefsm.bsky.social and @saxelab.bsky.social finally out!
We apply the statistical physics toolbox to analyse a simple model of reinforcement learning, and find some cool effects, like a speed-accuracy trade-off for generalisation 🚀
We apply the statistical physics toolbox to analyse a simple model of reinforcement learning, and find some cool effects, like a speed-accuracy trade-off for generalisation 🚀
Reposted by Sebastian Goldt
New paper with @leonlufkin.bsky.social and @eringrant.bsky.social!
Why do we see localized receptive fields so often, even in models without sparisity regularization?
We present a theory in the minimal setting from @aingrosso.bsky.social and @sebgoldt.bsky.social
Why do we see localized receptive fields so often, even in models without sparisity regularization?
We present a theory in the minimal setting from @aingrosso.bsky.social and @sebgoldt.bsky.social
December 13, 2024 at 10:49 AM
New paper with @leonlufkin.bsky.social and @eringrant.bsky.social!
Why do we see localized receptive fields so often, even in models without sparisity regularization?
We present a theory in the minimal setting from @aingrosso.bsky.social and @sebgoldt.bsky.social
Why do we see localized receptive fields so often, even in models without sparisity regularization?
We present a theory in the minimal setting from @aingrosso.bsky.social and @sebgoldt.bsky.social
Really interesting work from @leonlufkin.bsky.social, @saxelab.bsky.social and @eringrant.bsky.social !
December 13, 2024 at 10:01 AM
Really interesting work from @leonlufkin.bsky.social, @saxelab.bsky.social and @eringrant.bsky.social !
Reposted by Sebastian Goldt
If you missed it at the #NeurIPS2024 posters! Work led by @leonlufkin.bsky.social on analytical dynamics of localization in simple neural nets, as seen in real+artificial nets and distilled by @aingrosso.bsky.social @sebgoldt.bsky.social.
Leon is a fantastic collaborator + looking for PhD positions!
Leon is a fantastic collaborator + looking for PhD positions!
December 13, 2024 at 4:20 AM
If you missed it at the #NeurIPS2024 posters! Work led by @leonlufkin.bsky.social on analytical dynamics of localization in simple neural nets, as seen in real+artificial nets and distilled by @aingrosso.bsky.social @sebgoldt.bsky.social.
Leon is a fantastic collaborator + looking for PhD positions!
Leon is a fantastic collaborator + looking for PhD positions!
Reposted by Sebastian Goldt
Chronic Neuropixels friends - what's your trick to prevent tissue (re)growth which can make it difficult to retrieve the probe? Agarose? Mineral oil? Special craniotomy techniques? Any tip is welcome (NP1 and NP2 in mice with the NERF casing; van Daal et al. 2021)! #neuropixels
November 26, 2024 at 1:54 PM
Chronic Neuropixels friends - what's your trick to prevent tissue (re)growth which can make it difficult to retrieve the probe? Agarose? Mineral oil? Special craniotomy techniques? Any tip is welcome (NP1 and NP2 in mice with the NERF casing; van Daal et al. 2021)! #neuropixels
Better late than never...
November 25, 2024 at 7:33 AM
Better late than never...