Published September 13, 2025 | Version v1.0
Preprint Open

NeuroGlialNet: A Novel Brain-Inspired Architecture with Multi-Layer Astrocyte Networks

Authors/Creators

  • 1. Cohere Labs Community
  • 2. spairally.com

Description

We introduce NeuroGlialNet (NGN), a lightweight neural architecture that models the computational partnership between neurons and astrocytes in the brain. The architecture integrates multi-scale cortical processing with hierarchical astrocyte calcium dynamics through reaction-diffusion PDEs, spatial memory mechanisms, and quantum interference modulation. This concept paper presents the mathematical foundations and architectural innovations that enable brain-like computation in artificial systems. The current implementation has  ~ 17,000 parameters (~ 0.066 MB for 32-bit floats), corresponding to a model size of less than 0.1 MB .

Files

NeuroGlialNetv1.0.pdf

Files (275.9 kB)

Name Size Download all
md5:2d986a3929e4ad4e79ed9cefd14fddd2
275.9 kB Preview Download

Additional details

References

  • De Pittà, M., & Berry, H. (2019). Computational Glioscience. Springer
  • Araque, A., Carmignoto, G., Haydon, P. G., Oliet, S. H., Robitaille, R., & Volterra, A. (2014). Gliotransmitters travel in time and space. Neuron, 81(4), 728–739
  • . Chen, L., Zhang, Y., & Wang, M. (2024). Neural-glial hybrid architectures for brain-inspired AI: A comprehensive review. Nature Machine Intelligence, 6(3), 245–262
  • Rodriguez, P., Kim, S., & Thompson, R. (2024). Quantum-inspired modulation in artificial astrocyte networks. Science Advances, 10, eabk4821
  • . Liu, X., Patel, N., & Anderson, K. (2025). Multi-scale calcium dynamics in computational neuroglia: From theory to implementation. Cell, 188(4), 892–907
  • Onyango, W. F. (2025). QIMCNN : Quantum Interference Modulation Convolutional Neural Network (v1.0). Zenodo. https://doi.org/10.5281/zenodo.17018543