The Dynamics of Memory Replay, Diffusion and Attractor Formation
IEEE ICACI 2025 — China · 2025Read on IEEE ↗
Memory SystemsAttractor NetworksDiffusionNeuroscience
A study of how memory replay, semantic diffusion, and attractor formation interact to stabilize long-term representations in biological and artificial memory systems.
Abstract
This paper investigates the interplay between memory replay, diffusion processes, and attractor formation in recurrent memory systems. We propose a unified framework where replay acts as a stabilizing force counteracting semantic diffusion, leading to the emergence of robust attractor states that encode long-term knowledge.
Key Themes
- Replay dynamics as energy descent in latent memory manifolds
- Diffusion processes governing semantic drift and generalization
- Attractor formation as a mechanism for long-term knowledge consolidation
Venue
Presented at the IEEE International Conference on Advanced Computational Intelligence (ICACI) 2025, China.