Macroscopic fluctuations emerge in balanced networks with incomplete recurrent alignment
Itamar D. Landau, Haim Sompolinsky
Abstract
The authors generalize the theory of excitation-inhibition balance to recurrent networks with arbitrary low-rank connectivity structure and show incomplete alignment between the row-space and column-space of that connectivity leads to an emergence of macroscopic, low-dimensional fluctuations.
Topics & Concepts
Computer scienceMathematicsComplete informationBalance (ability)Sequence (biology)Statistical physicsComplex networkClass (philosophy)Artificial intelligenceNetwork structureAlgorithmComputationNeural dynamics and brain functionFunctional Brain Connectivity StudiesNonlinear Dynamics and Pattern Formation