Relation-Aware Dual-View Graph Contrastive Learning with Huber Covariance Whitening
Open Access
Abstract
Self-supervised graph collaborative filtering suffers from two geometric pathologies. In Dimensional Collapse, embeddings quietly collapse into a low-rank subspace, a well-documented but poorly solved problem. Semantic Collapse is more complex: stiff -squared orthogonalization penalties that are supposed to push the embeddings apart end up ripping through the heavy-tailed community overlaps that contain the collaborative signal. Earlier studies attempted to mitigate sparsity by injecting static noise or structural perturbations, but such interventions did not pinpoint the root cause, i.e., the distortions in the global covariance geometry itself. In this paper, we propose a Huber-Contrastive Graph Convolutional Network (HCGCN) that combines a spatial message-passing backbone with an contrastive augmentation overhead and a Relation-Aware Dual-View Gated Contrastive Network. The main novelty is a Huber Covariance Whitening module that imposes a geometry-aware threshold on the cross-correlation matrix of augmented views—below the threshold, the gradients follow an penalty (enforcing uniformity); above it, the penalty flattens to (protecting genuine semantic clusters from gradient explosion). This theoretically motivated dual-regime penalty actively preserves the macro-semantic topology of the graph while aggressively stamping out spurious noise correlations. The HCGCN is evaluated on the Yelp2018, Amazon-Book, and MovieLens datasets and performs significantly better than state-of-the-art baselines like LightGCN, SGL, SimGCL, and NESCL, especially under severe cold-start settings where covariance regulation proves most critical.
Keywords
This publication has 23 references indexed in Scilit:
- Acoustic signals-based probabilistic fault diagnosis for expansion joints of small and medium bridges using Bayesian ensemble learningEngineering Structures, 2026
- Multi-objective optimisation of surveillance camera placement for bridge–ship collision early-warning using an improved non-dominated sorting genetic algorithmAdvanced Engineering Informatics, 2025
- Neighborhood-Enhanced Supervised Contrastive Learning for Collaborative FilteringIEEE Transactions on Knowledge and Data Engineering, 2023
- Self-Supervised Learning for Recommender Systems: A SurveyIEEE Transactions on Knowledge and Data Engineering, 2023
- Bias and Debias in Recommender System: A Survey and Future DirectionsACM Transactions on Information Systems, 2023
- On Feature Decorrelation in Self-Supervised Learning2021
- Momentum Contrast for Unsupervised Visual Representation Learning2020
- T-GCN: A Temporal Graph Convolutional Network for Traffic PredictionIEEE Transactions on Intelligent Transportation Systems, 2019
- Matrix Factorization Techniques for Recommender SystemsComputer, 2009
- Robust Estimation of a Location ParameterThe Annals of Mathematical Statistics, 1964