Neural Particle Automata: Learning Self-Organizing Particle Dynamics
Summary
This Show HN presents Neural Particle Automata (NPA), a particle-based generalization of Neural Cellular Automata that operates on dynamic particle systems using differentiable SPH perception. It shows how local interactions can be learned with a shared neural rule, enabling robust self-organization, morphogenesis tasks, and texture synthesis, with scalable training via CUDA. The work highlights potential for new AI-driven simulations and graphics workflows, with open-source code on GitHub.