CoMA-DiT improves brain state decoding via cross-modal diffusion
September 11, 2026
CoMA-DiT uses a bidirectional Diffusion Transformer for latent augmentation in multimodal brain state decoding. By treating paired modalities as mutual generative supervision, it achieved accuracy gains of up to 4.28% in auditory attention decoding.
HOW THIS AFFECTS YOU
●
researcherYou can apply this cross-modal generative approach to enhance multimodal representation learning.
●
healthThis method could improve the precision of neurodiagnostic tools through better signal decoding.