new preprint, neural modeling survey, SfN2026, & more
Please help us understand what the field needs by filling out this brief survey. Your insights are essential in helping us develop tools that better serve the neuroscience community.
|
|
|
|
preprint: deep-learning assisted simulation of cortical circuit |
In our recent preprint, we introduce software that uses deep-learning tools to train a biologically realistic network of about 67,000 neurons representing the mouse visual cortex in roughly 6.5 hours on a single GPU. Trained with only simple moving patterns as visual stimuli, the network reproduced many features of real brain activity and responded realistically to new inputs, including natural scenes. The results show that preserving biological details is important not only for matching neural activity, but also for revealing how connections between neurons may give rise to cortical function.
|
|
|
|
DPointNet: a BMTK module for training bio-realistic networks |
Want to incorporate the methods for training bio-realistic networks, like in our recent preprint, into your own models and simulations? We have released the work as the latest component, called DPointNet, of the Brain Modeling Toolkit. It extends and generalizes the results of the paper to allow for a wider range of models, inputs, and training regimens – while also keeping the optimized efficiency. For instructions on how to use DPointNet, please see our newly added tutorials.
|
|
|
|
2026 Allen Institute Modeling Software Workshop |
|
|
|
In July, 26 participants joined 8 instructors for the 2026 Modeling Software Workshop, a three-day program focused on hands-on training with BMTK, SONATA, and VND tools. We are grateful to everyone who participated and helped create a welcoming, collaborative learning environment. Workshop tutorials and materials are now available.
|
|
| |
|
|
Are you coming to SfN2026? We will be there in Washington, DC, November 14-18. Find us at our posters or at the Allen Institute Booth!
|
-
PSTR208: "Simulations of cortical circuits across the cortical hierarchy" by Green et al.
- Monday, Nov. 16 | 1:00 p.m. – 5:00 p.m.
- PSTR230: “Cell-type- and layer-specific mechanisms of visually induced oscillations in a bio-realistic mouse V1 model” by Herrera et al.
- Tuesday, Nov. 17 | 8:00 a.m. – 12:00 p.m.
- PSTR291: "Deep-learning-assisted simulation of a cortical circuit: biological constraints as selectors of functionally equivalent solutions" by Ito et al.
- Tuesday, Nov. 17 | 1:00 p.m. – 5:00 p.m.
|
|
|
|
Shanahan Foundation/ applications now open |
|
|
|
|
The Shanahan Foundation Fellowship at the Interface of Data and Neuroscience is once again accepting applications. This unique opportunity invites new Ph.D.s with strong quantitative backgrounds to apply their expertise to our neuroscience datasets. Fellows will spend three years at the Allen Institute and University of Washington, with freedom to explore neuroscience through interdisciplinary collaboration while developing their own research projects. Applications for the Fall 2027 cohort are due December 1.
Visit the program page for details and application link.
| |
|
|
|