NHF v1: Hypergraph Fractal Neuron
Year: 2026
Type: article
Abstract: This deposit publishes NHF v1 (Hypergraph Fractal Neuron) as a cohesive, citation-ready technical paper and source package. NHF v1 proposes a modular alternative to the classical “sum-and-activate” neuron by introducing: meta-dendrites implemented as Micro-Processing Networks (MPP), a structured messaging interface via Multidimensional Information Packets (PIM), an internal integrated state (EAU), and an enriched firing representation (PD) transmitted through synapses treated as communication channels with an explicit Decoding Model (MD). A central goal of this release is terminological and structural coherence. The paper includes an intentionally extensive glossary, consistent notation across sections, and a taxonomy of potential uses to help readers evaluate where NHF can provide measurable value. The formalization is presented through two complementary routes: (A) a classical-math framing (information theory, tensor interactions, dynamical considerations, multiscale compression), and (B) an optional compact operator layer to shorten expressions once the architecture is understood. This v1 release focuses on making the concept implementable and testable rather than making broad performance claims. It emphasizes operational evaluation lenses (e.g., energy/latency proxies, calibration and uncertainty propagation, continual learning behavior, interface robustness) and highlights limitations and risks, including overhead and stability considerations when combining synaptic, structural, and topological plasticity. Included files NHF_v1_paper.pdf — compiled paper (primary artifact) NHF_v1_latex_source.zip — LaTeX source for reproducibility and future versions
Author: Queiroz, Carlos
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7118257142
FWCI:
Cites:
Cited by:
Related to: