Cross-Subject Neural Alignment via Soft Graph Matching and Dynamic Latent Space Projection for Brain-Computer Interfaces

Abstract

Individual variability in neural activation patterns presents a major challenge for non-invasive and invasive Brain-Computer Interfaces (BCIs). Direct cross-subject alignment often fails because identical sensory stimuli or cognitive tasks can evoke distinct neural representations across individuals. In this work, we propose a mathematical framework that leverages topological invariance to align neural responses into a universal latent space Z. By relaxing Quadratic Assignment Problems (QAP) to doubly stochastic soft graph matching, incorporating dynamic time-series confidence metrics, and optimizing a joint contrastive-topological loss function, our framework enables universal neural decoding without sacrificing individual relational brain structures.

Download full text (PDF)

Details

Type

Preprint

Published

October 31, 2026

Repository

Zenodo

DOI

10.5281/zenodo.23842524 (opens in a new tab)

Keywords

Brain-Computer Interfaces, Cross-Subject Neural Alignment, Soft Graph Matching, Latent Space Projection, Neural Representation Learning

Cite this work

APA

Kirdemir, T. A. (2026). Cross-Subject Neural Alignment via Soft Graph Matching and Dynamic Latent Space Projection for Brain-Computer Interfaces. Zenodo. https://doi.org/10.5281/zenodo.23842524

BibTeX

@misc{kirdemir2026,
  author    = {Kirdemir, Taha Arda},
  title     = {Cross-Subject Neural Alignment via Soft Graph Matching and Dynamic Latent Space Projection for Brain-Computer Interfaces},
  year      = {2026},
  month     = {10},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.23842524},
  url       = {https://doi.org/10.5281/zenodo.23842524},
  note      = {Preprint}
}