Incorporating GAN for Negative Sampling in Knowledge Representation Learning

• DOI: 10.1609/aaai.v32i1.11536
• Publication Date: 2022-06-24T21:33:45Z
• Authors: Peifeng Wang; Shuangyin Li; Rong Pan
• Keywords: Discriminator; Feature Learning; Representation; Margin (machine learning); Knowledge graph

Knowledge representation learning aims at modeling knowledge graph by encoding entities and relations into a low dimensional space. Most of the traditional works for embedding need negative sampling to minimize margin-based ranking loss. However, those construct samples through random mode, which are often too trivial fit model efficiently. In this paper, we propose novel framework based on Generative Adversarial Networks (GAN). GAN-based framework, take advantage generator obtain high-quality samples. Meanwhile, discriminator in GAN learns embeddings graph. Thus, can incorporate proposed various models improve ability learning. Experimental results show that our outperforms baselines triplets classification link prediction tasks.