Reproduce “Polyhedral Geometry of Time-to-First-Spike Neural Networks” 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
Manjot Singh, Guido Montúfar, Gitta Kutyniok
The authors analyze causal regions in time-to-first-spike networks, where a causal set contains presynaptic spikes arriving before a neuron fires. They show how these sets partition input space and how depth can multiply patterns through a constructive lower bound. In a three-input example, the SNN has 19 tuples versus at most 4 ReLU regions, under the stated conditions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.