Abstract
Multi-modal knowledge graph completion (MMKGC) aims to predict the missing triples in the multi-modal knowledge graphs by incorporating structural, visual, and textual information of entities into the discriminant models. The information from different modalities will work together to measure the triple plausibility. Existing MMKGC methods overlook the imbalance problem of modality information among entities, resulting in inadequate modal fusion and inefficient utilization of the raw modality information. To address the mentioned problems, we propose Adaptive Multi-modal Fusion and Modality Adversarial Training (AdaMF-MAT) to unleash the power of imbalanced modality information for MMKGC. AdaMF-MAT achieves multi-modal fusion with adaptive modality weights and further generates adversarial samples by modality-adversarial training to enhance the imbalanced modality information. Our approach is a co-design of the MMKGC model and training strategy which can outperform 19 recent MMKGC methods and achieve new state-of-the-art results on three public MMKGC benchmarks. Our code and data have been released at this https URL.
Abstract (translated)
多模态知识图 completion (MMKGC) 的目标是通过将实体的结构性、视觉和文本信息整合到判别模型中,预测多模态知识图中缺失的三元组。不同模态的信息将协同工作来衡量三元组的可能性。现有的MMKGC方法忽视了实体之间模态信息的不平衡问题,导致模态融合不充分,原始模态信息的利用率不高。为了解决这些问题,我们提出了自适应多模态融合和模态对抗训练(AdaMF-MAT)方法,以释放不平衡模态信息的潜力。AdaMF-MAT通过自适应的模态权重实现多模态融合,并通过模态对抗训练生成增强的不平衡模态信息。我们的方法是MMKGC模型和训练策略的联合设计,可以超越19个最近期的MMKGC方法,并在三个公共MMKGC基准上实现最先进的成果。我们的代码和数据已发布在https://这个网址。
URL
https://arxiv.org/abs/2402.15444