Network Embedding as Matrix Factorization: Unifying DeepWalk, LINE, PTE, and node2vec
论文地址:https://arxiv.org/abs/1710.02971
简介:
本文由微软和清华合作,在2018年发表,文章将DeepWalk, LINE, PTE, and node2vec算法在MF框架下进行了统一,在此基础上提出了NetMF算法
背景:
各种network embedding models很流行,但是对这些方法背后的理论分析却很少,之前的一些工作有:
Omer Levy and Yoav Goldberg. 2014. Neural Word Embedding as Implicit Matrix Factorization. In NIPS. 2177–218
(解释了skip-gram中负采样是对word-content矩阵的隐式分解)
Sanjeev Arora, Yuanzhi Li, Yingyu Liang, Tengyu Ma, and Andrej Risteski. 2016. A latent variable model approach to pmi-based word embeddings. TACL 4 (2016), 385–399
Tatsunori B Hashimoto, David Alvarez-Melis, and Tommi S Jaakkola. 2016. Word embeddings as metric recovery in semantic spaces. TACL 4 (2016), 273–286
(解释了 the word embedding m