Co-Modality Imbalanced Graph Contrastive Learning for Imbalenced Node Classification
Nov 28, 2022
Speakers
About
Graph contrastive learning (GCL), leveraging graph augmentations to convert graphs into different views and further train graph neural networks (GNNs), has achieved considerable success on graph benchmark datasets. Yet, there are still some gaps in directly applying existing GCL methods to real-world data. First, handcrafted graph augmentations require expert knowledge as well as trials and errors, but still can not yield consistent performance on multiple tasks. Second, most real-world graph data present imbalanced distribution but existing GCL methods are not immune to data imbalance. Therefore, this work proposes to explicitly tackle these challenges, via a principled framework called Co-Modality Imbalanced Graph Contratsive Learning with Network Pruning (CMI-GCL) to automatically generate contrastive pairs without expert knowledge and further learn balanced representation over unlabeled data. Specifically, we design inter-modality GCL to automatically generate contrastive pairs (e.g., node-image) based on node content. Inspired by the fact that minority samples can be “forgotten” by pruning deep neural networks, we naturally extend the ad-hoc compression technique, network pruning, to our GCL framework for detecting minority nodes. Based on this, we co-train two pruned encoders (e.g., GNN and image encoder) in different modalities by pushing the corresponding node-image pairs together and irrelevant node-image pairs away. Meanwhile, we also propose intra-modality GCL by co-training non-pruned GNN and pruned GNN, to ensure node embeddings with similar attributed features stay closed. By applying pre-trained CMI-GCL to two fine-tuning modes, we demonstrate that our model significantly outperforms state-of-the-art baseline models and learns more balanced representations on real-world graph datasets.Graph contrastive learning (GCL), leveraging graph augmentations to convert graphs into different views and further train graph neural networks (GNNs), has achieved considerable success on graph benchmark datasets. Yet, there are still some gaps in directly applying existing GCL methods to real-world data. First, handcrafted graph augmentations require expert knowledge as well as trials and errors, but still can not yield consistent performance on multiple tasks. Second, most real-world graph da…
Organizer
NeurIPS 2022
Account · 962 followers
Like the format? Trust SlidesLive to capture your next event!
Professional recording and live streaming, delivered globally.
Sharing
Recommended Videos
Presentations on similar topic, category or speaker