KBGAN: Adversarial Learning for Knowledge Graph Embeddings
This work addresses a key bottleneck in training knowledge graph embeddings for researchers and practitioners in AI, offering a generalizable method to enhance existing models, though it is incremental as it builds on prior embedding techniques and GANs.
The paper tackles the problem of generating effective negative training examples for knowledge graph embedding models, which typically only have positive facts, by introducing KBGAN, an adversarial learning framework that uses one embedding model as a negative sample generator to improve another as a discriminator, resulting in substantial performance improvements on link prediction tasks across datasets like FB15k-237, WN18, and WN18RR.
We introduce KBGAN, an adversarial learning framework to improve the performances of a wide range of existing knowledge graph embedding models. Because knowledge graphs typically only contain positive facts, sampling useful negative training examples is a non-trivial task. Replacing the head or tail entity of a fact with a uniformly randomly selected entity is a conventional method for generating negative facts, but the majority of the generated negative facts can be easily discriminated from positive facts, and will contribute little towards the training. Inspired by generative adversarial networks (GANs), we use one knowledge graph embedding model as a negative sample generator to assist the training of our desired model, which acts as the discriminator in GANs. This framework is independent of the concrete form of generator and discriminator, and therefore can utilize a wide variety of knowledge graph embedding models as its building blocks. In experiments, we adversarially train two translation-based models, TransE and TransD, each with assistance from one of the two probability-based models, DistMult and ComplEx. We evaluate the performances of KBGAN on the link prediction task, using three knowledge base completion datasets: FB15k-237, WN18 and WN18RR. Experimental results show that adversarial training substantially improves the performances of target embedding models under various settings.