DyGRAIN: An Incremental Learning Framework for Dynamic Graphs

Lee, Kyung-Min; Lee, Hyeongkeun

doi:10.24963/ijcai.2022/438

Cited by 4 publications

(3 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…MSCGL (Cai et al 2022) is designed for multimodal graphs with neural architectural search. DyGRAIN (Kim, Yun, and Kang 2022) explores the adaptation of receptive fields while distilling knowledge. Architectural methods modify the neural architecture of graph model itself, such as FGN (Wang et al 2022).…”

Section: Continual Graph Learningmentioning

confidence: 99%

“…In general, it aims at gradually learning new knowledge without catastrophic forgetting the old ones across sequentially coming tasks. Centered around fighting with forgetting, a series of methods (Kim, Yun, and Kang 2022;Galke et al 2021) have been proposed recently. Despite the success of prior works, continual graph learning still faces tremendous challenges.…”

Section: Introductionmentioning

confidence: 99%

“…, TWP(Liu, Yang, and Wang 2021), HPN(Zhang, Song, and Tao 2022), FGN(Wang et al 2022), MSCGL(Cai et al 2022) and DyGRAIN(Kim, Yun, and Kang 2022). ERGNN, TWP and DyGRAIN are implemented with GAT backbone(Veličković et al 2018).…”

mentioning

confidence: 99%

See 2 more Smart Citations

Self-Supervised Continual Graph Learning in Adaptive Riemannian Spaces

Peng

et al. 2023

AAAI

View full text Add to dashboard Cite

Continual graph learning routinely finds its role in a variety of real-world applications where the graph data with different tasks come sequentially. Despite the success of prior works, it still faces great challenges. On the one hand, existing methods work with the zero-curvature Euclidean space, and largely ignore the fact that curvature varies over the com- ing graph sequence. On the other hand, continual learners in the literature rely on abundant labels, but labeling graph in practice is particularly hard especially for the continuously emerging graphs on-the-fly. To address the aforementioned challenges, we propose to explore a challenging yet practical problem, the self-supervised continual graph learning in adaptive Riemannian spaces. In this paper, we propose a novel self-supervised Riemannian Graph Continual Learner (RieGrace). In RieGrace, we first design an Adaptive Riemannian GCN (AdaRGCN), a unified GCN coupled with a neural curvature adapter, so that Riemannian space is shaped by the learnt curvature adaptive to each graph. Then, we present a Label-free Lorentz Distillation approach, in which we create teacher-student AdaRGCN for the graph sequence. The student successively performs intra-distillation from itself and inter-distillation from the teacher so as to consolidate knowledge without catastrophic forgetting. In particular, we propose a theoretically grounded Generalized Lorentz Projection for the contrastive distillation in Riemannian space. Extensive experiments on the benchmark datasets show the superiority of RieGrace, and additionally, we investigate on how curvature changes over the graph sequence.

show abstract

Section: Continual Graph Learningmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%