Cross-Domain Few-Shot Graph Classification

We study the problem of few-shot graph classification across domains with nonequivalent feature spaces by introducing three new cross-domain benchmarks constructed from publicly available datasets. also propose an attention-based encoder that uses congruent views graphs, one contextual and two topological views, to learn representations task-specific information for fast adaptation, task-agnostic knowledge transfer. run exhaustive experiments evaluate performance contrastive meta-learning strategies. show when coupled metric-based frameworks, proposed achieves best average meta-test accuracy all benchmarks.