<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="liu-etal-2026-structural">
    <titleInfo>
        <title>Structural Anchor Pruning: Training-Free Multi-Vector Compression for Visual Document Retrieval</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Zhuchenyang</namePart>
        <namePart type="family">Liu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ziyu</namePart>
        <namePart type="family">Hu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yao</namePart>
        <namePart type="family">Zhang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yu</namePart>
        <namePart type="family">Xiao</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>Recent Vision-Language Models (e.g., ColPali) enable fine-grained Visual Document Retrieval (VDR) but incur prohibitive multi-vector index storage overhead. Existing training-free pruning methods either rely on heuristic layer choices or degrade sharply under aggressive compression, leading prior work to argue that effective high-compression pruning requires query-dependent training. We challenge this view with Structural Anchor Pruning (SAP), a self-calibrating, training-free, query-agnostic index-time framework combining (i) Score Retention (SR), a white-box per-layer compression diagnostic; (ii) SR-guided window selection, which automatically locates the structural pruning region of any backbone with no per-model hyperparameters; and (iii) a visual in-degree centrality scorer that identifies anchor patches within that window. On ViDoRe v1/v2 across three architectures spanning 18, 28, and 36 backbone layers, SAP retains 93–96% of NDCG@5 on v1 and 88–90% on the harder v2 while pruning 90% of visual tokens; at 20× compression it retains 85–90% and 76–79% respectively. Our layer-resolved SR analysis reveals an Alignment-Aggregation Divergence: visual structure is preserved as a stable “Structural Plateau” within the backbone, while the final layers reshape it into a sparse, query-aligned form unsuitable for pruning. Probing the pre-retrieval base backbones shows that contrastive fine-tuning sharpens this boundary three- to eight-fold, explaining why final-layer methods fail.</abstract>
    <identifier type="citekey">liu-etal-2026-structural</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.216/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>4947</start>
            <end>4975</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Structural Anchor Pruning: Training-Free Multi-Vector Compression for Visual Document Retrieval
%A Liu, Zhuchenyang
%A Hu, Ziyu
%A Zhang, Yao
%A Xiao, Yu
%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 liu-etal-2026-structural
%X Recent Vision-Language Models (e.g., ColPali) enable fine-grained Visual Document Retrieval (VDR) but incur prohibitive multi-vector index storage overhead. Existing training-free pruning methods either rely on heuristic layer choices or degrade sharply under aggressive compression, leading prior work to argue that effective high-compression pruning requires query-dependent training. We challenge this view with Structural Anchor Pruning (SAP), a self-calibrating, training-free, query-agnostic index-time framework combining (i) Score Retention (SR), a white-box per-layer compression diagnostic; (ii) SR-guided window selection, which automatically locates the structural pruning region of any backbone with no per-model hyperparameters; and (iii) a visual in-degree centrality scorer that identifies anchor patches within that window. On ViDoRe v1/v2 across three architectures spanning 18, 28, and 36 backbone layers, SAP retains 93–96% of NDCG@5 on v1 and 88–90% on the harder v2 while pruning 90% of visual tokens; at 20× compression it retains 85–90% and 76–79% respectively. Our layer-resolved SR analysis reveals an Alignment-Aggregation Divergence: visual structure is preserved as a stable “Structural Plateau” within the backbone, while the final layers reshape it into a sparse, query-aligned form unsuitable for pruning. Probing the pre-retrieval base backbones shows that contrastive fine-tuning sharpens this boundary three- to eight-fold, explaining why final-layer methods fail.
%U https://aclanthology.org/2026.emnlp-main.216/
%P 4947-4975

Markdown (Informal)

[Structural Anchor Pruning: Training-Free Multi-Vector Compression for Visual Document Retrieval](https://aclanthology.org/2026.emnlp-main.216/) (Liu et al., EMNLP 2026)

• Structural Anchor Pruning: Training-Free Multi-Vector Compression for Visual Document Retrieval (Liu et al., EMNLP 2026)

ACL

• Zhuchenyang Liu, Ziyu Hu, Yao Zhang, and Yu Xiao. 2026. Structural Anchor Pruning: Training-Free Multi-Vector Compression for Visual Document Retrieval. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 4947–4975, Budapest, Hungary. Association for Computational Linguistics.