contrastive learning (CL) has emerged as a dominant technique for
unsupervised representation learning which embeds augmented versions of the
anchor close to each other (positive samples) and pushes the embeddings of
other samples (negatives) apart. As revealed in recent studies, CL ca
对比学习在图学习领域吸引了大量研究兴趣,现有的图对比学习方法需要大规模和多样化的负样本来确保嵌入的质量,但这样会引入虚假的负样本,同时增加了计算负担和时间复杂度,为解决这些问题,提出了一种简单而有效的模型 GraphRank,通过引入基于排名的学习来测量相似度得分,成功地缓解了虚假负样本问题,并将时间复杂度从 O (N^2) 降低到 O (N),并在多个图任务上展开了广泛实验,证明 GraphRank 在各种任务中表现优异。