English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Advances in optical and electrophysiological recording technologies have madeit possible to record the dynamics of thousands of neurons, opening up newpossibilities for interpreting and controlling large neural populations inbehaving animals. A promising way to extract computational principles fromthese large datasets is to train data-constrained recurrent neural networks(dRNNs). Performing this training in real-time could open doors for researchtechniques and medical applications to model and control interventions atsingle-cell resolution and drive desired forms of animal behavior. However,existing training algorithms for dRNNs are inefficient and have limitedscalability, making it a challenge to analyze large neural recordings even inoffline scenarios. To address these issues, we introduce a training methodtermed Convex Optimization of Recurrent Neural Networks (CORNN). In studies ofsimulated recordings, CORNN attained training speeds ~100-fold faster thantraditional optimization approaches while maintaining or enhancing modelingaccuracy. We further validated CORNN on simulations with thousands of cellsthat performed simple computations such as those of a 3-bit flip-flop or theexecution of a timed response. Finally, we showed that CORNN can robustlyreproduce network dynamics and underlying attractor structures despitemismatches between generator and inference models, severe subsampling ofobserved neurons, or mismatches in neural time-scales. Overall, by trainingdRNNs with millions of parameters in subminute processing times on a standardcomputer, CORNN constitutes a first step towards real-time network reproductionconstrained on large-scale neural recordings and a powerful computational toolfor advancing the understanding of neural computation.

CORNN: Convex optimization of recurrent neural networks for rapid inference of neural dynamics