FRACTAL MEMORY ON A HYPERSPHERE: Hybrid Optical Neuron Architecture for Extreme Data Compression
Year: 2025
Type: article
Abstract: FRACTAL MEMORY ON A HYPERSPHERE represents an emerging paradigm in neuromorphic computing, inspired by the self-similar structures of fractal geometry. This article presents a hybrid hardware architecture based on an optical neuron with polarization filters and directional chromatic photodiodes, integrated onto a physical hypersphere with atomic platinum layers. Utilizing double stereographic projections on the Riemann sphere and quaternion extensions for 4D rotations, the system enables efficient hierarchical storage with lossless compression on the order of 1000:1. Numerical simulations validate the model, demonstrating a 45% reduction in interferences and an SNR of ~25 dB at 1 GHz. Applications target edge AI, computer vision, and cognitive modeling, paving the way for adaptive low-power chips.
Author: PREDA, NELU
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7103136629
FWCI:
Cites:
Cited by:
Related to: