arXiv：cs.LG· Benjamin Loire, Galadriel Bri\`ere, C\'elia Brahimi, Antoine Toffano, Ana\"is Baudot·· 3 小时前AI 评分35
KGATE：一个知识图谱嵌入训练环境
KGATE : a Knowledge Graph Embedding Training Environment
AI 导读
研究者发布 KGATE（Knowledge Graph Autoencoder Training Environment），一个基于 PyTorch Geometric 和 TorchKGE 构建的模块化 Python 库，支持将初始化器、编码器、解码器、损失函数、负采样器和评估指标作为组件自由组装。
正文
Abstract:Knowledge graph embedding (KGE) models encode the entities and relations of a knowledge graph into a low-dimensional latent space, enabling tasks such as classification or link prediction. Most KGE models follow an autoencoder architecture, in which an encoder projects the knowledge graph into the latent space and a decoder reconstruct it. Combining both encoder and decoder components is increasingly needed, yet existing libraries rarely support complete autoencoders, are often unmaintained, rely on undocumented default hyperparameters, and produce results that cannot be compared across libraries. Here we present KGATE (Knowledge Graph Autoencoder Training Environment), a modular Python library built on PyTorch Geometric and TorchKGE. KGATE lets users assemble initializers, encoders, decoders, losses, negative samplers, and evaluation metrics as building blocks, or plug in their own block. KGATE includes a preprocessing procedure that controls data leakage, a builtin training pipeline, and reproducibility by design. Benchmarks against six existing KGE libraries show that KGATE training time is comparable with the fastest libraries while offering a broader set of features.
| Comments: | Main paper (7 pages, 1 figure) and supplementary materials (4 pages, 1 figure, 3 tables) provided |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.09927 [cs.LG] |
| (or arXiv:2610.09927v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09927 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Benjamin Loire [view email]
[v1]
Wed, 7 Oct 2026 12:14:50 UTC (455 KB)
来源：arXiv：cs.LG · arxiv.org