Improving Dropout in Graph Convolutional Networks for Recommendation via Contrastive Loss.
Hiroki OkamuraKeisuke MaedaRen TogoTakahiro OgawaMiki HaseyamaPublished in: ICASSP (2023)
Keyphrases
- average degree
- social networks
- heterogeneous social networks
- highly connected
- small world
- recommender systems
- network structure
- collaborative filtering
- graph structure
- fully connected
- degree distribution
- overlapping communities
- graph theoretic
- random walk
- graph representation
- graph theory
- citation networks
- community discovery
- dynamic networks
- deep learning
- betweenness centrality
- graph layout
- complex networks
- graph structures
- directed graph
- random graphs
- path length
- user preferences
- heterogeneous networks
- edge weights
- directed acyclic graph
- weighted graph
- graphical models
- connected components