Reproduce “Embedding of Low-Dimensional Sensory Dynamics in Recurrent Networks: Implications for the Geometry of Neural Representation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Vikas N. O'Reilly-Shah, Alessandro Maria Selvitella
The paper explains how a recurrent network driven by a sensory signal can develop a ring or torus that tracks the signal. Under contraction and a smoothness condition, generic networks with hidden dimension N>2d can embed sensory dynamics. A theorem links prediction to scale-limited state separation. Simulations found 0.065 and 0.029; they do not establish biology or causality.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.