CTLN

Welcome to the CTLN Project Website!
This project is funded by NIH R01 EB022862 (2016-19).
PI: Carina Curto (Penn State)
co-I: Katherine Morrison (U. Northern Colorado)
Postdocs: Jesse Geneson & Chris Langdon (Penn State)
Other collaborators: Anda Degeratu, Vladimir Itskov, Samantha Moore, Caitlyn Parmelee, Eva Pastalkova, David Rolnick

CTLN stands for Combinatorial Threshold-Linear Network. This is a recurrent attractor neural network model, whose dynamics are shaped by the structure of an underlying connectivity graph. Unlike the classical Hopfield model, CTLN networks are not typically symmetric and feature both static and dynamic attractors.

This website is under construction. Just a few items so far…

0. NEW! Our math preprint, on fixed points of competitive TLNs, where we prove all the graph rules for CTLNs. A new version is available, accepted to Neural Computation.

1. Slides from Carina’s Delaware talk (Sept 2018): Delaware-talk-sep29-xsxseo

2. Slides from Carina’s Cosyne talk (Mar 2018): Cosyne slides.

3. Our book chapter on graphical rules for CTLNs, and the corresponding Matlab code.

4. Our (somewhat outdated) 2016 preprint on the CTLN model, and accompanying Matlab code.

5. Our poster for the 2016 BRAIN Initiative PI meeting: CTLN poster.

6. Listen to our network songs!

7. Slides from Carina’s talk in Sapporo, Japan (Aug 2017): Japan slides.

8. Slides from Katie’s ICMNS talk in Boulder, CO (Jun 2017): ICMNS slides.