MEG-to-MEG Transfer Learning and Cross-Task Speech/Silence Detection with Limited Data

Xabier, Vincenzo, Eva, Ibon, Mathieu, Nicola · 2026

Read MEG-to-MEG Transfer Learning and Cross-Task Speech/Silence Detection with Limited Data by Xabier, Vincenzo, Eva, Ibon, Mathieu, Nicola (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-18253.txt.

Description

Xabier de Zuazo and five co-authors show brain-speech decoding can travel across people and tasks. Their 2026 paper tackles data hunger in speech brain-computer interfaces built on MEG recordings. The team pretrains a Conformer model on fifty hours of listening data from one person, then fine-tunes on just five minutes per person across eighteen participants.

Transfer lifts scores within each task by one to four percent, and across tasks by up to five or six percent. Crucially, models trained on speech production still decode passive listening above chance, evidence that the learned patterns reflect shared neural processes rather than mere motor activity. The paper gives the first demonstration of MEG transfer learning and cross-task decoding spanning perception and production, a step toward practical speech interfaces.

Catalog record

CREATORXabier, Vincenzo, Eva, Ibon, Mathieu, Nicola
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE22 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-18253

Browse the full library · About the Tower of Babel