Geodesic Flow Matching for Denoising High-Dimensional Structured Representations
Quick Take
The study introduces Geodesic Flow Matching to enhance denoising in high-dimensional Spatial Semantic Pointers (SSPs) by adhering to their toroidal manifold structure. This method reduces tracking error by 72% and improves neural efficiency by 40% in a Spiking Neural SLAM system, outperforming traditional Euclidean approaches.
Key Points
- Geodesic Flow Matching adapts Riemannian dynamics for SSP denoising.
- Traditional Flow Matching fails due to Euclidean assumptions on SSPs.
- Achieves 72% reduction in tracking error in Spiking Neural SLAM.
- Increases neural efficiency by 40% compared to competitive methods.
- Code available at https://github.com/kremHabashy/CleanupSSP.
Article Content
From source RSS / original summaryarXiv:2606. 00248v1 Announce Type: new Abstract: Vector Symbolic Algebras (VSAs) enable robust neurosymbolic reasoning by encoding symbolic information into high-dimensional distributed representations. For continuous domains, Spatial Semantic Pointers (SSPs) extend this framework by mapping variables onto continuous toroidal manifolds. However, standard approaches like Flow Matching assume a flat Euclidean geometry, which fails to account for the geometric constraints imposed on valid SSP states.
We demonstrate that this assumption fails for SSPs: Euclidean linear interpolants ``cut through" the manifold's interior, destroying the phase and magnitude structure required for accurate decoding. To resolve this, we employ Geodesic Flow Matching, adapting Riemannian transport dynamics to strictly restrict the denoising flow to the SSP toroidal manifold. We validate this approach in a Spiking Neural SLAM system, showing that manifold-aware cleanup stabilizes path integration against drift.
The method achieves a 72\% reduction in tracking error and enables a 40\% increase in neural efficiency compared to competitive baselines. Code is available at https://github. com/kremHabashy/CleanupSSP .
Reader Mode unavailable (could not extract clean content).
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →MindGames Arena Generalization Track: In2AI Solution with Delayed Per-Step Reward Attribution
The In2AI solution introduces delayed per-step reward attribution for training language model agents in multi-agent environments, achieving top performance on the MindGames Arena benchmark at NeurIPS 2025. An 8-billion-parameter model outperformed larger proprietary systems, including GPT-5, in competitive play, demonstrating enhanced stability and sample efficiency in reinforcement learning.
