@inproceedings{gordeev-etal-2025-hypercomplex,
title = "Hypercomplex Transformer: Novel Attention Mechanism",
author = "Gordeev, Maxim and
Aleksandr, Zuev and
Bakulin, Mikhail and
Latyshev, Andrey and
Kozlov, Dmitry and
Yao, Yiwu and
Anastasia, Voronova",
editor = "Inui, Kentaro and
Sakti, Sakriani and
Wang, Haofen and
Wong, Derek F. and
Bhattacharyya, Pushpak and
Banerjee, Biplab and
Ekbal, Asif and
Chakraborty, Tanmoy and
Singh, Dhirendra Pratap",
booktitle = "Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics",
month = dec,
year = "2025",
address = "Mumbai, India",
publisher = "The Asian Federation of Natural Language Processing and The Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-ijcnlp.115/",
pages = "1845--1851",
ISBN = "979-8-89176-303-6",
abstract = "Self-attention mechanisms have become foundational across modern deep learning architectures. Recent efforts focus on improving their efficiency, particularly for signal processing tasks. The existing approaches employ complex-valued representations for inputs and weights and achieve higher accuracy at the cost of increased model size and inference latency. Dual-numbered algebra offers a promising alternative that allows efficient multiplication and faster inference with the same representational capacity. Inspired by previous studies in the field of hypercomplex neural networks, we introduce a generalized hypercomplex attention block and integrate it into Transformer-based models for EEG classification. Our experiments include adaptation of the hypercomplex models, so that the number of parameters is equal to that of their real-valued counterparts. Across all scenarios, the dual- and complex-numbered models consistently outperform the real ones, demonstrating superior accuracy. This work presents hypercomplex attention as a competitive and computationally efficient strategy with potential value to solve multiple NLP tasks."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gordeev-etal-2025-hypercomplex">
<titleInfo>
<title>Hypercomplex Transformer: Novel Attention Mechanism</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maxim</namePart>
<namePart type="family">Gordeev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zuev</namePart>
<namePart type="family">Aleksandr</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mikhail</namePart>
<namePart type="family">Bakulin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Andrey</namePart>
<namePart type="family">Latyshev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dmitry</namePart>
<namePart type="family">Kozlov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yiwu</namePart>
<namePart type="family">Yao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Voronova</namePart>
<namePart type="family">Anastasia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kentaro</namePart>
<namePart type="family">Inui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sakriani</namePart>
<namePart type="family">Sakti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haofen</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Derek</namePart>
<namePart type="given">F</namePart>
<namePart type="family">Wong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pushpak</namePart>
<namePart type="family">Bhattacharyya</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Biplab</namePart>
<namePart type="family">Banerjee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Asif</namePart>
<namePart type="family">Ekbal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dhirendra</namePart>
<namePart type="given">Pratap</namePart>
<namePart type="family">Singh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>The Asian Federation of Natural Language Processing and The Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Mumbai, India</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-303-6</identifier>
</relatedItem>
<abstract>Self-attention mechanisms have become foundational across modern deep learning architectures. Recent efforts focus on improving their efficiency, particularly for signal processing tasks. The existing approaches employ complex-valued representations for inputs and weights and achieve higher accuracy at the cost of increased model size and inference latency. Dual-numbered algebra offers a promising alternative that allows efficient multiplication and faster inference with the same representational capacity. Inspired by previous studies in the field of hypercomplex neural networks, we introduce a generalized hypercomplex attention block and integrate it into Transformer-based models for EEG classification. Our experiments include adaptation of the hypercomplex models, so that the number of parameters is equal to that of their real-valued counterparts. Across all scenarios, the dual- and complex-numbered models consistently outperform the real ones, demonstrating superior accuracy. This work presents hypercomplex attention as a competitive and computationally efficient strategy with potential value to solve multiple NLP tasks.</abstract>
<identifier type="citekey">gordeev-etal-2025-hypercomplex</identifier>
<location>
<url>https://aclanthology.org/2025.findings-ijcnlp.115/</url>
</location>
<part>
<date>2025-12</date>
<extent unit="page">
<start>1845</start>
<end>1851</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Hypercomplex Transformer: Novel Attention Mechanism
%A Gordeev, Maxim
%A Aleksandr, Zuev
%A Bakulin, Mikhail
%A Latyshev, Andrey
%A Kozlov, Dmitry
%A Yao, Yiwu
%A Anastasia, Voronova
%Y Inui, Kentaro
%Y Sakti, Sakriani
%Y Wang, Haofen
%Y Wong, Derek F.
%Y Bhattacharyya, Pushpak
%Y Banerjee, Biplab
%Y Ekbal, Asif
%Y Chakraborty, Tanmoy
%Y Singh, Dhirendra Pratap
%S Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics
%D 2025
%8 December
%I The Asian Federation of Natural Language Processing and The Association for Computational Linguistics
%C Mumbai, India
%@ 979-8-89176-303-6
%F gordeev-etal-2025-hypercomplex
%X Self-attention mechanisms have become foundational across modern deep learning architectures. Recent efforts focus on improving their efficiency, particularly for signal processing tasks. The existing approaches employ complex-valued representations for inputs and weights and achieve higher accuracy at the cost of increased model size and inference latency. Dual-numbered algebra offers a promising alternative that allows efficient multiplication and faster inference with the same representational capacity. Inspired by previous studies in the field of hypercomplex neural networks, we introduce a generalized hypercomplex attention block and integrate it into Transformer-based models for EEG classification. Our experiments include adaptation of the hypercomplex models, so that the number of parameters is equal to that of their real-valued counterparts. Across all scenarios, the dual- and complex-numbered models consistently outperform the real ones, demonstrating superior accuracy. This work presents hypercomplex attention as a competitive and computationally efficient strategy with potential value to solve multiple NLP tasks.
%U https://aclanthology.org/2025.findings-ijcnlp.115/
%P 1845-1851
Markdown (Informal)
[Hypercomplex Transformer: Novel Attention Mechanism](https://aclanthology.org/2025.findings-ijcnlp.115/) (Gordeev et al., Findings 2025)
ACL
- Maxim Gordeev, Zuev Aleksandr, Mikhail Bakulin, Andrey Latyshev, Dmitry Kozlov, Yiwu Yao, and Voronova Anastasia. 2025. Hypercomplex Transformer: Novel Attention Mechanism. In Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics, pages 1845–1851, Mumbai, India. The Asian Federation of Natural Language Processing and The Association for Computational Linguistics.