Chae, Tony, Avramidis, Kleanthis, Emmanouilidou, Dimitra, Tashev, Ivan J. · Oct 26, 2026
Authors present MtReNIRS for transferable neurovascular representations
Markdown · JSON · Original source
The authors Tony Chae, Kleanthis Avramidis, Dimitra Emmanouilidou and Ivan J. Tashev present MtReNIRS, a model for functional near-infrared spectroscopy (fNIRS). They say it was on 12 public datasets covering 268 subjects and 567 recording hours, then evaluated for transfer across three external datasets. A montage-aware masked autoencoder is a model that learns from partially hidden signal data while accounting for differences in sensor layouts, or montages. The authors say MtReNIRS maps heterogeneous montages onto a shared anatomical reference and encodes fNIRS signals, chromophore identity and cortical location. They report cross-subject performance on par with or better than fully supervised baselines, and better results than prior state-of-the-art fNIRS models in within-subject calibration. They also report robustness to sparse montages. Escape Meridian interpretation: If those results transfer beyond the evaluated datasets, the approach could reduce the amount of labeled data needed to build brain-computer interfaces; the abstract does not establish real-world performance or deployment. The paper argues that montage-aware self-supervision may offer a scalable foundation for label-efficient brain-computer interfaces. How well does MtReNIRS transfer to new devices, tasks and users outside the three external datasets?
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
q-bio.NC · Article
Source details
Original source date: Oct 26, 2026 (2026-10-26).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-05T02:31:24.319Z. First observed: 2026-10-28T04:04:34.230Z. Retrieved: 2026-10-29T19:03:29.199Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The evidence supplied is the abstract; it does not provide performance metrics, evaluation details, or evidence of real-world deployment. The abstract reports transfer across three external datasets but does not establish performance on other devices, tasks or users.
Open question
How well does MtReNIRS transfer to new devices, tasks and users outside the three external datasets?
Source-stated
The authors report pretraining MtReNIRS on 12 public fNIRS datasets spanning 268 subjects and 567 recording hours.
pretrained on 12 public fNIRS datasets spanning 268 subjects and 567 recording hours
Source-stated
The model transfers across three external datasets, according to the authors.
learns generalizable neurovascular representations that transfer across three external datasets
Source-stated
The authors say MtReNIRS maps heterogeneous montages onto a shared anatomical reference and encodes fNIRS signals, chromophore identity and cortical location.
jointly encodes the slow hemodynamic nature of fNIRS, chromophore identity and cortical location by mapping heterogeneous montages onto a shared anatomical reference
Source-stated
The authors report cross-subject performance on par with or better than fully supervised baselines, and better few-shot within-subject calibration than prior state-of-the-art fNIRS models.
achieves cross-subject performance on par or better than fully supervised baselines and excels in few-shot within-subject calibration, where it outperforms prior state-of-the-art fNIRS models
Escape Meridian interpretation
If the reported results transfer beyond the evaluated datasets, the method could reduce the labeled data needed for brain-computer interfaces; this implication is not established by the abstract.
providing a scalable foundation for label-efficient brain-computer interfaces
Source and contributor context
Chae, Tony, Avramidis, Kleanthis, Emmanouilidou, Dimitra, Tashev, Ivan J. · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Pre-training
A model's initial training on a large dataset, before further adaptation for particular uses.
Zero-shot
Doing a task without worked examples (zero-shot) or with only a few examples given in the prompt (few-shot).
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: 309c8bf5-adf8-4224-802e-041ac92e4b6b.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-05T02:31:24.319Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: 37d63b5edf0c9772dc4efc356523081dec0ea973761d8c34c0288e6fbd0d1edd. Source text hash: b6da1c9bdc3ca442d6527cc4d6d811dd6a44b1e8b050a2f6f4de4978ca8bbb32.
Revision and source-check history
Version 1 · 2026-11-05T02:31:24.319Z
Authors present MtReNIRS for transferable neurovascular representations
The authors Tony Chae, Kleanthis Avramidis, Dimitra Emmanouilidou and Ivan J. Tashev present MtReNIRS, a model for functional near-infrared spectroscopy (fNIRS). They say it was on 12 public datasets covering 268 subjects and 567 recording hours, then evaluated for transfer across three external datasets. A montage-aware masked autoencoder is a model that learns from partially hidden signal data while accounting for differences in sensor layouts, or montages. The authors say MtReNIRS maps heterogeneous montages onto a shared anatomical reference and encodes fNIRS signals, chromophore identity and cortical location. They report cross-subject performance on par with or better than fully supervised baselines, and better results than prior state-of-the-art fNIRS models in within-subject calibration. They also report robustness to sparse montages. Escape Meridian interpretation: If those results transfer beyond the evaluated datasets, the approach could reduce the amount of labeled data needed to build brain-computer interfaces; the abstract does not establish real-world performance or deployment. The paper argues that montage-aware self-supervision may offer a scalable foundation for label-efficient brain-computer interfaces. How well does MtReNIRS transfer to new devices, tasks and users outside the three external datasets?
Original source
q-bio.NC · Article
Source details
Original source date: Oct 26, 2026 (2026-10-26).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-05T02:31:24.319Z. First observed: 2026-10-28T04:04:34.230Z. Retrieved: 2026-10-29T19:03:29.199Z.
Limits
The evidence supplied is the abstract; it does not provide performance metrics, evaluation details, or evidence of real-world deployment. The abstract reports transfer across three external datasets but does not establish performance on other devices, tasks or users.
Open question
How well does MtReNIRS transfer to new devices, tasks and users outside the three external datasets?
Source-stated
The authors report pretraining MtReNIRS on 12 public fNIRS datasets spanning 268 subjects and 567 recording hours.
pretrained on 12 public fNIRS datasets spanning 268 subjects and 567 recording hours
Source-stated
The model transfers across three external datasets, according to the authors.
learns generalizable neurovascular representations that transfer across three external datasets
Source-stated
The authors say MtReNIRS maps heterogeneous montages onto a shared anatomical reference and encodes fNIRS signals, chromophore identity and cortical location.
jointly encodes the slow hemodynamic nature of fNIRS, chromophore identity and cortical location by mapping heterogeneous montages onto a shared anatomical reference
Source-stated
The authors report cross-subject performance on par with or better than fully supervised baselines, and better few-shot within-subject calibration than prior state-of-the-art fNIRS models.
achieves cross-subject performance on par or better than fully supervised baselines and excels in few-shot within-subject calibration, where it outperforms prior state-of-the-art fNIRS models
Escape Meridian interpretation
If the reported results transfer beyond the evaluated datasets, the method could reduce the labeled data needed for brain-computer interfaces; this implication is not established by the abstract.
providing a scalable foundation for label-efficient brain-computer interfaces
Source and contributor context
Chae, Tony, Avramidis, Kleanthis, Emmanouilidou, Dimitra, Tashev, Ivan J. · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Pre-training
A model's initial training on a large dataset, before further adaptation for particular uses.
Zero-shot
Doing a task without worked examples (zero-shot) or with only a few examples given in the prompt (few-shot).
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: 309c8bf5-adf8-4224-802e-041ac92e4b6b.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-05T02:31:24.319Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: 37d63b5edf0c9772dc4efc356523081dec0ea973761d8c34c0288e6fbd0d1edd. Source text hash: b6da1c9bdc3ca442d6527cc4d6d811dd6a44b1e8b050a2f6f4de4978ca8bbb32.
uncertain · 2026-11-05T02:31:24.319Z · anthropic/claude-opus-5.5
Publication: 309c8bf5-adf8-4224-802e-041ac92e4b6b
attribution — pass: The four named authors match the arXiv listing. Performance and transfer claims are consistently framed as the authors' reports, and the conditional inference is labeled as Escape Meridian interpretation.
Authors:Tony Chae, Kleanthis Avramidis, Dimitra Emmanouilidou, Ivan J. Tashev
summary — uncertain: The central term fNIRS is expanded but never explained in plain language. A lay reader is not told it uses light to track blood-flow changes linked to brain activity. Other central terms are also left unexplained: chromophore identity, cortical location, few-shot within-subject calibration, fully supervised baselines and foundation model. The draft says the model encodes fNIRS signals, while the source says it encodes their slow hemodynamic nature. The question is also repeated verbatim at the end of the summary. Smallest repair: add brief plain-language glosses or tooltips for these terms and restore the hemodynamic wording.
jointly encodes the slow hemodynamic nature of fNIRS, chromophore identity and cortical location
date — pass: The originalPublished date of 2026-10-26 matches the v1 arXiv submission date. It is not the later feed date.
[Submitted on 26 Oct 2026]
support — pass: Pretraining scale, transfer, montage mapping and comparative results are all stated in the abstract. The brain-computer-interface implication is explicitly conditional and notes that the abstract does not establish real-world performance. The question is open, not leading. The limitations accurately note that no metrics are provided.
providing a scalable foundation for label-efficient brain-computer interfaces
media — pass: The selected media is the verified direct arXiv abstract page for this paper, which is the most direct access to the source.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
A model's initial training on a large dataset, before further adaptation for particular uses.
Definition referenceDoing a task without worked examples (zero-shot) or with only a few examples given in the prompt (few-shot).
Definition referencearXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivA model's initial training on a large dataset, before further adaptation for particular uses.
Definition referenceDoing a task without worked examples (zero-shot) or with only a few examples given in the prompt (few-shot).
Definition referencearXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXiv