Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204686239> ?p ?o ?g. }
- W3204686239 abstract "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 (negative samples) apart. As revealed in recent works, CL can benefit from hard negative samples (negative samples that are difficult to distinguish from the anchor). However, we observe minor improvement or even performance drop when we adopt existing hard negative mining techniques in Graph Contrastive Learning (GCL). We find that many hard negative samples similar to anchor point are false negative ones (samples from the same class as anchor point) in GCL, which is different from CL in computer vision and will lead to unsatisfactory performance of existing hard negative mining techniques in GCL. To eliminate this bias, we propose Debiased Graph Contrastive Learning (DGCL), a novel and effective method to estimate the probability whether each negative sample is true or not. With this probability, we devise two schemes (i.e., DGCL-weight and DGCL-mix) to boost the performance of GCL. Empirically, DGCL outperforms or matches previous unsupervised state-of-the-art results on several benchmarks and even exceeds the performance of supervised ones." @default.
- W3204686239 created "2021-10-11" @default.
- W3204686239 creator A5021345474 @default.
- W3204686239 creator A5041519364 @default.
- W3204686239 creator A5051700680 @default.
- W3204686239 creator A5082786719 @default.
- W3204686239 creator A5087672753 @default.
- W3204686239 date "2021-10-05" @default.
- W3204686239 modified "2023-09-27" @default.
- W3204686239 title "Debiased Graph Contrastive Learning." @default.
- W3204686239 cites W2169012810 @default.
- W3204686239 cites W2194775991 @default.
- W3204686239 cites W2519887557 @default.
- W3204686239 cites W2554952599 @default.
- W3204686239 cites W2621048175 @default.
- W3204686239 cites W2624431344 @default.
- W3204686239 cites W2766453196 @default.
- W3204686239 cites W2786915849 @default.
- W3204686239 cites W2798991696 @default.
- W3204686239 cites W2951451004 @default.
- W3204686239 cites W2961295589 @default.
- W3204686239 cites W2962756421 @default.
- W3204686239 cites W2963782635 @default.
- W3204686239 cites W2964420626 @default.
- W3204686239 cites W3012816161 @default.
- W3204686239 cites W3013713234 @default.
- W3204686239 cites W3033039844 @default.
- W3204686239 cites W3034978746 @default.
- W3204686239 cites W3035524453 @default.
- W3204686239 cites W3039465721 @default.
- W3204686239 cites W3039806961 @default.
- W3204686239 cites W306662213 @default.
- W3204686239 cites W3081963674 @default.
- W3204686239 cites W3094468028 @default.
- W3204686239 cites W3100292948 @default.
- W3204686239 cites W3102419180 @default.
- W3204686239 cites W3103995645 @default.
- W3204686239 cites W3104097132 @default.
- W3204686239 cites W3106485021 @default.
- W3204686239 cites W3121904315 @default.
- W3204686239 cites W3123939835 @default.
- W3204686239 cites W3141587922 @default.
- W3204686239 cites W3156636935 @default.
- W3204686239 cites W3174415560 @default.
- W3204686239 cites W3204453541 @default.
- W3204686239 cites W3125808012 @default.
- W3204686239 hasPublicationYear "2021" @default.
- W3204686239 type Work @default.
- W3204686239 sameAs 3204686239 @default.
- W3204686239 citedByCount "0" @default.
- W3204686239 crossrefType "posted-content" @default.
- W3204686239 hasAuthorship W3204686239A5021345474 @default.
- W3204686239 hasAuthorship W3204686239A5041519364 @default.
- W3204686239 hasAuthorship W3204686239A5051700680 @default.
- W3204686239 hasAuthorship W3204686239A5082786719 @default.
- W3204686239 hasAuthorship W3204686239A5087672753 @default.
- W3204686239 hasConcept C119857082 @default.
- W3204686239 hasConcept C132525143 @default.
- W3204686239 hasConcept C153180895 @default.
- W3204686239 hasConcept C154945302 @default.
- W3204686239 hasConcept C2524010 @default.
- W3204686239 hasConcept C28719098 @default.
- W3204686239 hasConcept C33923547 @default.
- W3204686239 hasConcept C41008148 @default.
- W3204686239 hasConcept C80444323 @default.
- W3204686239 hasConceptScore W3204686239C119857082 @default.
- W3204686239 hasConceptScore W3204686239C132525143 @default.
- W3204686239 hasConceptScore W3204686239C153180895 @default.
- W3204686239 hasConceptScore W3204686239C154945302 @default.
- W3204686239 hasConceptScore W3204686239C2524010 @default.
- W3204686239 hasConceptScore W3204686239C28719098 @default.
- W3204686239 hasConceptScore W3204686239C33923547 @default.
- W3204686239 hasConceptScore W3204686239C41008148 @default.
- W3204686239 hasConceptScore W3204686239C80444323 @default.
- W3204686239 hasLocation W32046862391 @default.
- W3204686239 hasOpenAccess W3204686239 @default.
- W3204686239 hasPrimaryLocation W32046862391 @default.
- W3204686239 hasRelatedWork W1642428186 @default.
- W3204686239 hasRelatedWork W171557998 @default.
- W3204686239 hasRelatedWork W2810874585 @default.
- W3204686239 hasRelatedWork W2902200315 @default.
- W3204686239 hasRelatedWork W2942123544 @default.
- W3204686239 hasRelatedWork W2950352605 @default.
- W3204686239 hasRelatedWork W2953137836 @default.
- W3204686239 hasRelatedWork W2963691377 @default.
- W3204686239 hasRelatedWork W2968414212 @default.
- W3204686239 hasRelatedWork W3035161783 @default.
- W3204686239 hasRelatedWork W3047171579 @default.
- W3204686239 hasRelatedWork W3097564492 @default.
- W3204686239 hasRelatedWork W3106328333 @default.
- W3204686239 hasRelatedWork W3109894131 @default.
- W3204686239 hasRelatedWork W3126925884 @default.
- W3204686239 hasRelatedWork W3185678184 @default.
- W3204686239 hasRelatedWork W3196575433 @default.
- W3204686239 hasRelatedWork W3197112563 @default.
- W3204686239 hasRelatedWork W3203064196 @default.
- W3204686239 hasRelatedWork W3211356784 @default.
- W3204686239 isParatext "false" @default.
- W3204686239 isRetracted "false" @default.
- W3204686239 magId "3204686239" @default.