<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="tamiti-etal-2026-udss">
    <titleInfo>
        <title>UDSS-BWE: Uncertainty- and Decision-Science Inspired Swin BandWidth Extension</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Tarikul</namePart>
        <namePart type="given">Islam</namePart>
        <namePart type="family">Tamiti</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Sajid</namePart>
        <namePart type="given">Fardin</namePart>
        <namePart type="family">Dipto</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">David</namePart>
        <namePart type="family">Vergano</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Luke</namePart>
        <namePart type="family">Baja-Ricketts</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Anomadarshi</namePart>
        <namePart type="family">Barua</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
    </relatedItem>
    <abstract>Bandwidth extension (BWE) is fundamentally localized: the most perceptual distortions are not average-case distortions, but rare high-frequency (HF) transients that standard, risk-neutral objectives tend to smooth away. To close this gap, we seek solutions in the risk-sensitive and uncertainty-aware decision science rules and present UDSS-BWE, which introduces five decision-science and uncertainty-aware dicriminators: CVaRD (does tail pooling to amplify HF artifacts), CCD (a primal-dual augmented Lagrangian to prevent HF overboost), MCUD (a learnable utility over spectral flatness/ centroid/ rolloff), EDD (captures epistemic uncertainty), and DROD (captures entropic KL-DRO aggregation). UDSS-BWE is also designed as a complex-valued adversarial BWE framework that uses Swin-based generators, a lightweight dual-stream shifted-window backbone, to capture local and long-range structure efficiently, while learnable lattice coupling provides controlled cross-stream exchange. UDSS-BWE is optimized extensively and achieves better performance across a total of 13 different evaluation metrics with 3.89x fewer parameters (72M vs. 18M) over two English and French datasets under clean and noisy conditions. To best of our knowledge, this work shows how multi disciplinary decision-science and uncertainty theories can be successfully used to design efficient discriminators for producing more nuanced audios, establishing a new baseline in the BWE task.</abstract>
    <identifier type="citekey">tamiti-etal-2026-udss</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.2429/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>55142</start>
            <end>55175</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T UDSS-BWE: Uncertainty- and Decision-Science Inspired Swin BandWidth Extension
%A Tamiti, Tarikul Islam
%A Dipto, Sajid Fardin
%A Vergano, David
%A Baja-Ricketts, Luke
%A Barua, Anomadarshi
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F tamiti-etal-2026-udss
%X Bandwidth extension (BWE) is fundamentally localized: the most perceptual distortions are not average-case distortions, but rare high-frequency (HF) transients that standard, risk-neutral objectives tend to smooth away. To close this gap, we seek solutions in the risk-sensitive and uncertainty-aware decision science rules and present UDSS-BWE, which introduces five decision-science and uncertainty-aware dicriminators: CVaRD (does tail pooling to amplify HF artifacts), CCD (a primal-dual augmented Lagrangian to prevent HF overboost), MCUD (a learnable utility over spectral flatness/ centroid/ rolloff), EDD (captures epistemic uncertainty), and DROD (captures entropic KL-DRO aggregation). UDSS-BWE is also designed as a complex-valued adversarial BWE framework that uses Swin-based generators, a lightweight dual-stream shifted-window backbone, to capture local and long-range structure efficiently, while learnable lattice coupling provides controlled cross-stream exchange. UDSS-BWE is optimized extensively and achieves better performance across a total of 13 different evaluation metrics with 3.89x fewer parameters (72M vs. 18M) over two English and French datasets under clean and noisy conditions. To best of our knowledge, this work shows how multi disciplinary decision-science and uncertainty theories can be successfully used to design efficient discriminators for producing more nuanced audios, establishing a new baseline in the BWE task.
%U https://aclanthology.org/2026.emnlp-main.2429/
%P 55142-55175

Markdown (Informal)

[UDSS-BWE: Uncertainty- and Decision-Science Inspired Swin BandWidth Extension](https://aclanthology.org/2026.emnlp-main.2429/) (Tamiti et al., EMNLP 2026)

• UDSS-BWE: Uncertainty- and Decision-Science Inspired Swin BandWidth Extension (Tamiti et al., EMNLP 2026)

ACL

• Tarikul Islam Tamiti, Sajid Fardin Dipto, David Vergano, Luke Baja-Ricketts, and Anomadarshi Barua. 2026. UDSS-BWE: Uncertainty- and Decision-Science Inspired Swin BandWidth Extension. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 55142–55175, Budapest, Hungary. Association for Computational Linguistics.