
The proposed SE-HSSL framework introduces three sampling-efficient self-supervised signals - node-level CCA, group-level CCA, and hierarchical membership-level contrast - to effectively learn discriminative hypergraph representations without relying on arbitrary negative sampling.


coremsg

sampling-efficient-hypergraph-self-supervised-learning-with-tri-directional-signals


Sampling-Efficient Hypergraph Self-Supervised Learning with Tri-Directional Signals


title_rewrite


Convolutional Signal Propagation (CSP) is a simple and scalable algorithm for learning representations on hypergraphs, which can be applied to tasks such as node classification and retrieval.


convolutional-signal-propagation-a-simple-and-scalable-algorithm-for-hypergraph-representation-learning


Convolutional Signal Propagation: A Simple and Scalable Algorithm for Hypergraph Representation Learning
