K-ON stacks K prediction heads onto an LLM to generate entity tokens in one step and uses entity-level contrastive learning, achieving new state-of-the-art results on two knowledge graph completion benchmarks.
MEAformer: Multi-modal Entity Alignment Transformer for Meta Modality Hybrid
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Multi-modal entity alignment (MMEA) aims to discover identical entities across different knowledge graphs (KGs) whose entities are associated with relevant images. However, current MMEA algorithms rely on KG-level modality fusion strategies for multi-modal entity representation, which ignores the variations of modality preferences of different entities, thus compromising robustness against noise in modalities such as blurry images and relations. This paper introduces MEAformer, a multi-modal entity alignment transformer approach for meta modality hybrid, which dynamically predicts the mutual correlation coefficients among modalities for more fine-grained entity-level modality fusion and alignment. Experimental results demonstrate that our model not only achieves SOTA performance in multiple training scenarios, including supervised, unsupervised, iterative, and low-resource settings, but also has a limited number of parameters, efficient runtime, and interpretability. Our code is available at https://github.com/zjukg/MEAformer.
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
K-ON: Stacking Knowledge On the Head Layer of Large Language Model
K-ON stacks K prediction heads onto an LLM to generate entity tokens in one step and uses entity-level contrastive learning, achieving new state-of-the-art results on two knowledge graph completion benchmarks.