Saved in:
Bibliographic Details
Main Authors: Jhilal, Soufiane, Martin, Stéphanie, Giraud, Anne-Lise
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.15909
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908781649068032
author Jhilal, Soufiane
Martin, Stéphanie
Giraud, Anne-Lise
author_facet Jhilal, Soufiane
Martin, Stéphanie
Giraud, Anne-Lise
contents Non-invasive decoding of imagined speech remains challenging due to weak, distributed signals and limited labeled data. Our paper introduces an image-based approach that transforms magnetoencephalography (MEG) signals into time-frequency representations compatible with pretrained vision models. MEG data from 21 participants performing imagined speech tasks were projected into three spatial scalogram mixtures via a learnable sensor-space convolution, producing compact image-like inputs for ImageNet-pretrained vision architectures. These models outperformed classical and non-pretrained models, achieving up to 90.4% balanced accuracy for imagery vs. silence, 81.0% vs. silent reading, and 60.6% for vowel decoding. Cross-subject evaluation confirmed that pretrained models capture shared neural representations, and temporal analyses localized discriminative information to imagery-locked intervals. These findings show that pretrained vision models applied to image-based MEG representations can effectively capture the structure of imagined speech in non-invasive neural signals.
format Preprint
id arxiv_https___arxiv_org_abs_2601_15909
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Transfer Learning from ImageNet for MEG-Based Decoding of Imagined Speech
Jhilal, Soufiane
Martin, Stéphanie
Giraud, Anne-Lise
Computation and Language
Artificial Intelligence
Computer Vision and Pattern Recognition
Non-invasive decoding of imagined speech remains challenging due to weak, distributed signals and limited labeled data. Our paper introduces an image-based approach that transforms magnetoencephalography (MEG) signals into time-frequency representations compatible with pretrained vision models. MEG data from 21 participants performing imagined speech tasks were projected into three spatial scalogram mixtures via a learnable sensor-space convolution, producing compact image-like inputs for ImageNet-pretrained vision architectures. These models outperformed classical and non-pretrained models, achieving up to 90.4% balanced accuracy for imagery vs. silence, 81.0% vs. silent reading, and 60.6% for vowel decoding. Cross-subject evaluation confirmed that pretrained models capture shared neural representations, and temporal analyses localized discriminative information to imagery-locked intervals. These findings show that pretrained vision models applied to image-based MEG representations can effectively capture the structure of imagined speech in non-invasive neural signals.
title Transfer Learning from ImageNet for MEG-Based Decoding of Imagined Speech
topic Computation and Language
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2601.15909