Current browse context:
cs.LG
Change to browse by:
References & Citations
Computer Science > Machine Learning
Title: Highly Efficient Knowledge Graph Embedding Learning with Orthogonal Procrustes Analysis
(Submitted on 10 Apr 2021 (v1), last revised 17 Apr 2021 (this version, v2))
Abstract: Knowledge Graph Embeddings (KGEs) have been intensively explored in recent years due to their promise for a wide range of applications. However, existing studies focus on improving the final model performance without acknowledging the computational cost of the proposed approaches, in terms of execution time and environmental impact. This paper proposes a simple yet effective KGE framework which can reduce the training time and carbon footprint by orders of magnitudes compared with state-of-the-art approaches, while producing competitive performance. We highlight three technical innovations: full batch learning via relational matrices, closed-form Orthogonal Procrustes Analysis for KGEs, and non-negative-sampling training. In addition, as the first KGE method whose entity embeddings also store full relation information, our trained models encode rich semantics and are highly interpretable. Comprehensive experiments and ablation studies involving 13 strong baselines and two standard datasets verify the effectiveness and efficiency of our algorithm.
Submission history
From: Xutan Peng [view email][v1] Sat, 10 Apr 2021 03:55:45 GMT (2772kb,D)
[v2] Sat, 17 Apr 2021 12:17:05 GMT (2772kb,D)
Link back to: arXiv, form interface, contact.