Reproduce “Conformal Isometry of Lie Group Representation in Recurrent Network of Grid Cells” 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
Dehong Xu, Ruiqi Gao, Wen-Hao Zhang, Xue-Xin Wei, Ying Nian Wu
The authors train recurrent networks to encode position and update it from self-motion, constraining equal-length movements in different directions to produce equally large, orthogonal changes in population activity. The networks develop hexagon-like activity maps and integrate paths. In a 500-step test, a 20-step vanilla RNN has lower error when re-encoded every 20 steps; this does not establish a mechanism in animals.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.