Fitting state space model to RNNs does a really good job (capturing like 90% of the variance)
To understand RNN computations, we globally linearized their hidden-unit activity using high-dimensional linear-Gaussian state-space models (SSMs).
High-d SSMs (latents dim > obs dim) have great performance, and are interpretable through tools from dynamical systems and control theory.
How do brain areas control each other? 🧠🎛️
✨In our NeurIPS 2025 Spotlight paper, we introduce a data-driven framework to answer this question using deep learning, nonlinear control, and differential geometry.🧵⬇️
Title slide: distinct modes of generalization from parameters and context and two paths to bridge the gap
I was honored to speak at Princeton’s symposium on The Physics of John Hopfield: Learning & Intelligence this week. I sketched out a perspective that ties together some of our recent work on ICL vs. parametric learning, and some possible links to hippocampal replay: 1/
📢 The full programme for the 2nd Workshop on Biological Control Systems (WBCS 2025) is out!
🗓️ 19 Nov 2025 | Online across time zones
🔗 Programme & info: biocontrolseminars.org/workshop-2025
pcts.princeton.edu/events/2025/...
At the end of the talk addressing scaling they seem to indicate a role for their architecture as a Gibbs sampling pipeline/seed for a GAN running (deterministically) on GPU
extropic.ai
This talk on Extropic’s thermodynamic computing approach has a nice discussion of ML on GPUs versus their approach and the detail of the latter leveraging transistor noise for probability distribution sampling
youtu.be/dRuhl6MLC78
www.newton.ac.uk/event/scl/
www.birs.ca/events/2025/...
Bifurcations—an underexplored concept in neuroscience—can help explain how small differences in neural circuits give rise to entirely novel functions, writes Xiao-Jing Wang.
#neuroskyence
www.thetransmitter.org/neural-dynam...
Trick or treat?
🎙️ New episode! We travel back to 19th-century Russia to meet Lyapunov, the mathematician who gave science the language of stability. From his friendship with Chebyshev to a life marked by devotion and tragedy, his 1892 thesis still shapes how we understand dynamics, uncertainty, and control today.
@andykeller.bsky.social @kempnerinstitute.bsky.social presented “Flow Equivariant Cybernetics”, a blueprint for agents that learn through continuous feedback with their environment.
This year’s meeting will bring together leading researchers in #neuroscience, #robotics, #machinelearning, #controltheory, and #theoreticalneuroscience to explore how behaviour and intelligence emerge from complex feedback loops linking brains, bodies, and environments.
Great talk (form and content) to accompany the book:
youtu.be/gs38tMxZseA
🚦 New episode! Legendary traffic-control pioneer Markos Papageorgiou (Technical University of Crete) joins us to discuss how intelligent control algorithms help cut congestion, reduce CO₂ emissions, and keep cities moving for millions every day! 😮
📢Join us for our online seminar tomorrow Tuesday Aug 26 at 11am ET (5pm CET)
Frank Britto Bisso from Carnegie Mellon University will present "Programming genetic circuits that operate as neural networks"
As always more information on our website biocontrolseminars.org
Traveling waves exist and provide efficient solutions to fundamental computational problems. It seems likely that evolution embraced them.
doi.org/10.48550/arX...
#neuroscience
🎙️ New episode! Cosimo takes us inside the fascinating world of soft robotics 🤖🐙 Compliance over rigidity, shape as intelligence, and why underactuation may actually be a feature — unmissable! 😄
New preprint!
Why are long-range connectomic interactions in the cortex dominant in shaping dynamics in some experiments but apparently negligible in others?
We (w/ R Maran, @elimuller.bsky.social) address this question by studying a new hybrid model of cortical dynamics.
arxiv.org/abs/2506.19800
The detailed schedule for the #BernsteinConference is now available on our website. Check out the Invited Talks and Contributed Talks and start planning your conference visit! 👉 bernstein-network.de/bernstein-co...
#BernsteinNetwork #CompNeuro
cnd.northwestern.edu/neurocomputa...
Our piece on the emerging field of NeuroAI, with Claudia Clopath, in Nat. Rev. Neuroscience:
rdcu.be/ezLlY
🔄 Updated Arxiv Paper
Title: Koopman-based control of nonlinear systems with closed-loop guarantees
Authors: Robin Str\"asser, Julian Berberich, Manuel Schaller, Karl Worthmann, Frank Allg\"ower
Read more: https://arxiv.org/abs/2411.10359
Congratulations to Sara A. Solla for receiving the Valentin Braitenberg Award for #ComputationalNeuroscience 2025! 🎉
The award ceremony and accompanying lecture will take place during the #BernsteinConference.
👉 bit.ly/4okssfz
#BernsteinNetwork
If those get sticky the PC can read it as an extended press and treat that as an intended interrupt. If it’s a toggle switch ignore that remark
That’s all I got, good luck
