Reproduce “Low Tensor Rank Learning of Neural Dynamics” 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

Arthur Pellegrino, N Alex Cayco-Gajic, Angus Chadwick

The authors introduce a model that describes learning-related changes in recurrent network connectivity through a few shared patterns whose strengths vary across trials. In a reach simulation, the pretraining weight change was rank three and the perturbation-learning update tensor rank two. In macaque recordings, a five-component fit matched a full tensor model on held-out data, without identifying biological synapses.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.