SPD-MetaFormer is what you need for small-data brain decoding

Name: SPD-MetaFormer is what you need for small-data brain decoding
Uploaded: 2026-10-14T17:02:25.756115+00:00
Duration: 320 s
Description: SPD-MetaFormer is an attention-free architecture for brain decoding that uses uniform geometric aggregation under log-Euclidean geometry. It separates local memory from a global summary, refining the summary through uniform mixing. This method achieves competitive results in small-data brain decoding across EEG benchmarks, outperforming existing baselines.

SPD-MetaFormer is an attention-free architecture for brain decoding that uses uniform geometric aggregation under log-Euclidean geometry. It separates local memory from a global summary, refining the summary through uniform mixing. This method achieves competitive results in small-data brain decoding across EEG benchmarks, outperforming existing baselines.

Open the full episode