Lightnews — Scholar-powered news

A Erdem Sagtekin

@aesagtekin.bsky.social

620 followers 3 following 2 posts

chronic reader of old papers. dynamical systems & computation & bio-plausible learning. curr: @flatiron ccn, msc: comp neuro @tubingen, bsc: EE.

Posts Media Videos Starter Packs

Reposted by A Erdem Sagtekin

Friedemann Zenke @fzenke.bsky.social · May 27

1/6 Why does the brain maintain such precise excitatory-inhibitory balance?
Our new preprint explores a provocative idea: Small, targeted deviations from this balance may serve a purpose: to encode local error signals for learning.
www.biorxiv.org/content/10.1...
led by @jrbch.bsky.social

4 56 170

Reposted by A Erdem Sagtekin

Matthijs Pals @matthijspals.bsky.social · Dec 11

How to find all fixed points in piece-wise linear recurrent neural networks (RNNs)?
A short thread 🧵
 In RNNs with N units with ReLU(x-b) activations the phase space is partioned in 2^N regions by hyperplanes at x=b 1/7

1 12 60

Reposted by A Erdem Sagtekin

David G. Clark @david-g-clark.bsky.social · Dec 3

(1/5) Fun fact: Several classic results in the stat. mech. of learning can be derived in a couple lines of simple algebra!

In this paper with Haim Sompolinsky, we simplify and unify derivations for high-dimensional convex learning problems using a bipartite cavity method.
arxiv.org/abs/2412.01110

Simplified derivations for high-dimensional convex learning problems

Statistical physics provides tools for analyzing high-dimensional problems in machine learning and theoretical neuroscience. These calculations, particularly those using the replica method, often invo...

arxiv.org

2 16 55

A Erdem Sagtekin @aesagtekin.bsky.social · Nov 9

This list likely reflects mainly my interests and circle, and I’m sure I’ve missed many people, but I gave it a try: (I’ll be slowly editing it until it reaches 150/150)

go.bsky.app/7VFUkdn

(also, I tried but couldn't remove my profile...)

48 51 80

A Erdem Sagtekin @aesagtekin.bsky.social · Nov 1

i enjoyed reading the geometry of plasticity paper and felt that something important was coming, this is it:

Blake Richards @tyrellturing.bsky.social · Oct 28

Check out our new preprint! We show there is a better learning algorithm for #compneuro than gradient descent (GD): exponentiated gradients (EG).

tl;dr: EG respects Dale's law, produces weight distributions that match biology, and outperforms GD in biologically relevant scenarios.

🧠📈 🧪

Jonathan Cornford @repromancer.bsky.social · Oct 28

Why does #compneuro need new learning methods? ANN models are usually trained with Gradient Descent (GD), which violates biological realities like Dale’s law and log-normal weights. Here we describe a superior learning algorithm for comp neuro: Exponentiated Gradients (EG)! 1/12 #neuroscience 🧪

1 1 4