@inproceedings{wang-etal-2026-atoms,
title = "Atoms and Grammar of the Heart: Geometry-Aligned Discrete Tokenization for {ECG} Language Modeling",
author = "Wang, Zirui and
Yang, Qiuling and
Wu, Pang and
Wang, Peng and
Li, Zhenfeng and
Chen, Xianxiang and
Du, Lidong and
Fang, Zhen",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.685/",
pages = "14369--14393",
ISBN = "979-8-89176-462-0",
abstract = "Extending discrete tokenization and autoregressive modeling to signals whose latent states follow continuous trajectories introduces phase{--}morphology ambiguity. Similar local shapes can occur at distinct trajectory positions, while the same position can vary across instances. Standard vector quantization (VQ) may therefore merge contextually different patches and split corresponding ones. We trace this behavior to radial-to-tangential gradient leakage under the straight-through estimator and propose **Geometry Alignment Quantization (GAQ)**. GAQ combines a learned Mahalanobis reshaping of the codebook with a Householder backward map. The map exactly cancels the leakage in the idealized setting, while the numerical stabilization used in practice introduces a bounded deviation. We instantiate GAQ on ECG as **AtomECG**, a tokenizer of sub-waveform atoms composed by a GPT grammar. Atom sequences show long-context scaling from perplexity 488 to 24, split-replicated motif enrichment yields auditable clinical hypotheses, and AtomECG outperforms eight eSSL baselines on more than ten diagnostic tasks. A known-phase limit cycle and PPG results further support GAQ on phase-like signals beyond ECG."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2026-atoms">
<titleInfo>
<title>Atoms and Grammar of the Heart: Geometry-Aligned Discrete Tokenization for ECG Language Modeling</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zirui</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qiuling</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pang</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Peng</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhenfeng</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xianxiang</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lidong</namePart>
<namePart type="family">Du</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Fang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Extending discrete tokenization and autoregressive modeling to signals whose latent states follow continuous trajectories introduces phase–morphology ambiguity. Similar local shapes can occur at distinct trajectory positions, while the same position can vary across instances. Standard vector quantization (VQ) may therefore merge contextually different patches and split corresponding ones. We trace this behavior to radial-to-tangential gradient leakage under the straight-through estimator and propose **Geometry Alignment Quantization (GAQ)**. GAQ combines a learned Mahalanobis reshaping of the codebook with a Householder backward map. The map exactly cancels the leakage in the idealized setting, while the numerical stabilization used in practice introduces a bounded deviation. We instantiate GAQ on ECG as **AtomECG**, a tokenizer of sub-waveform atoms composed by a GPT grammar. Atom sequences show long-context scaling from perplexity 488 to 24, split-replicated motif enrichment yields auditable clinical hypotheses, and AtomECG outperforms eight eSSL baselines on more than ten diagnostic tasks. A known-phase limit cycle and PPG results further support GAQ on phase-like signals beyond ECG.</abstract>
<identifier type="citekey">wang-etal-2026-atoms</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.685/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>14369</start>
<end>14393</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Atoms and Grammar of the Heart: Geometry-Aligned Discrete Tokenization for ECG Language Modeling
%A Wang, Zirui
%A Yang, Qiuling
%A Wu, Pang
%A Wang, Peng
%A Li, Zhenfeng
%A Chen, Xianxiang
%A Du, Lidong
%A Fang, Zhen
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F wang-etal-2026-atoms
%X Extending discrete tokenization and autoregressive modeling to signals whose latent states follow continuous trajectories introduces phase–morphology ambiguity. Similar local shapes can occur at distinct trajectory positions, while the same position can vary across instances. Standard vector quantization (VQ) may therefore merge contextually different patches and split corresponding ones. We trace this behavior to radial-to-tangential gradient leakage under the straight-through estimator and propose **Geometry Alignment Quantization (GAQ)**. GAQ combines a learned Mahalanobis reshaping of the codebook with a Householder backward map. The map exactly cancels the leakage in the idealized setting, while the numerical stabilization used in practice introduces a bounded deviation. We instantiate GAQ on ECG as **AtomECG**, a tokenizer of sub-waveform atoms composed by a GPT grammar. Atom sequences show long-context scaling from perplexity 488 to 24, split-replicated motif enrichment yields auditable clinical hypotheses, and AtomECG outperforms eight eSSL baselines on more than ten diagnostic tasks. A known-phase limit cycle and PPG results further support GAQ on phase-like signals beyond ECG.
%U https://aclanthology.org/2026.findings-emnlp.685/
%P 14369-14393
Markdown (Informal)
[Atoms and Grammar of the Heart: Geometry-Aligned Discrete Tokenization for ECG Language Modeling](https://aclanthology.org/2026.findings-emnlp.685/) (Wang et al., Findings 2026)
ACL
- Zirui Wang, Qiuling Yang, Pang Wu, Peng Wang, Zhenfeng Li, Xianxiang Chen, Lidong Du, and Zhen Fang. 2026. Atoms and Grammar of the Heart: Geometry-Aligned Discrete Tokenization for ECG Language Modeling. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 14369–14393, Budapest, Hungary. Association for Computational Linguistics.