Reproduce “Predicate Neuron: Differentiable Quantization of Predicate Relations in Gradient-based Learning” 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
Ioannis Konstantoulas, Iliana Loi, Kyriakos Sgarbas
The Predicate Neuron adds a fuzzy-logic operation to neural networks, selecting among AND, OR, and XOR while using a smooth gradient surrogate. On a constructed three-branch discontinuous function, the Predicate Network reports validation MAE of 1.7830, compared with 93.8606 for a Dense network without a logic core. Dense controls with logic cores eventually reach similar validation loss, but more slowly.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.