Current studentsExisting staffContact us
 

Cluster-aware prompt ensemble learning for few-shot vision-language model adaptation

Simple item page
dc.contributor.authorChen, Z.
dc.contributor.authorYu, X.
dc.contributor.authorTao, X.
dc.contributor.authorLi, Y.
dc.contributor.authorHuang, Z.
dc.date.issued2026
dc.description.abstractVision-language models (VLMs) such as CLIP achieve zero-shot transfer across various tasks by pre-training on numerous image-text pairs. These models often benefit from using an ensemble of context prompts to represent a class. Despite being effective, conventional prompt ensembling that averages textual features of context prompts often yields suboptimal results. This is because feature averaging shifts the class centroids away from the true class distribution. To address this issue, we propose the Cluster-Aware Prompt Ensemble Learning (CAPEL) framework, which preserves the cluster nature of context prompts. CAPEL classifies images into one of several class clusters, each represented by a distinct prompt. Instead of ensembling prompts in the feature space, we perform ensembling in the classification logits space, aligning better with the visual feature distribution. To further optimize prompt fine-tuning while maintaining cluster-specific discriminative power, we introduce a cluster-preserving regularization term. This ensures that prompts remain distinct and specialized for different clusters, preventing collapse into a uniform direction. Additionally, we integrate an adaptive prompt weighting technique to dynamically adjust the attention weights for flawed or ambiguous prompts, ensuring robust performance across diverse datasets and tasks.
dc.description.statementofresponsibilityZhi Chen, Xin Yu, Xiaohui Tao, Yan Li, Zi Huang
dc.identifier.citationPattern Recognition, 2026; 172:112596-1-112596-13
dc.identifier.doi10.1016/j.patcog.2025.112596
dc.identifier.issn0031-3203
dc.identifier.issn1873-5142
dc.identifier.orcidYu, X. [0000-0001-9890-5489] [0000-0002-0269-5649] [0000-0002-3388-9606] [0000-0002-6265-9519]
dc.identifier.urihttps://hdl.handle.net/2440/149921
dc.language.isoen
dc.publisherElsevier
dc.relation.granthttp://purl.org/au-research/grants/arc/DP240101814
dc.rights© 2025 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
dc.source.urihttps://doi.org/10.1016/j.patcog.2025.112596
dc.subjectensemble learning; vision-language models; conditional entropy; logits ensemble
dc.titleCluster-aware prompt ensemble learning for few-shot vision-language model adaptation
dc.typeJournal article
pubs.publication-statusPublished

Files

Original bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
hd_149921.pdf
Size:
8.35 MB
Format:
Adobe Portable Document Format
Description:
Published version
Download

Collections

Research Outputs