GNN
Recommendation Systems (7): Graph Neural Networks and Social Recommendation
A deep, intuition-first walkthrough of graph neural networks for recommendation: GCN, GAT, GraphSAGE, PinSage, LightGCN, NGCF, social signals, scalable sampling, and cold start. Diagrams plus working PyTorch.
PDE and Machine Learning (8): Reaction-Diffusion Systems and Graph Neural Networks
Deep GNNs collapse because they are diffusion equations on graphs. Turing's reaction-diffusion theory tells us how to fix it -- and closes the eight-chapter PDE+ML series.
HCGR: Hyperbolic Contrastive Graph Representation Learning for Session-based Recommendation
HCGR embeds session graphs in the Lorentz model of hyperbolic space and trains them with InfoNCE-style contrastive learning. This review unpacks why hierarchical session intent fits hyperbolic geometry, how Lorentz …
paper2repo: GitHub Repository Recommendation for Academic Papers
paper2repo aligns academic papers with GitHub repositories in a shared embedding space using a constrained GCN. Covers the joint heterogeneous graph, the WARP ranking loss, the cosine alignment constraint, and the full …
Session-based Recommendation with Graph Neural Networks (SR-GNN)
SR-GNN turns a click session into a directed weighted graph and runs a gated GNN to predict the next item. Covers session-graph construction, GGNN updates, attention-based session pooling, training, benchmarks, and the …
Graph Contextualized Self-Attention Network (GC-SAN) for Session-based Recommendation
GC-SAN combines a session-graph GGNN (local transitions) with multi-layer self-attention (global dependencies) for session-based recommendation. Covers graph construction, message passing, attention fusion, and where the …
LLMGR: Integrating Large Language Models with Graphical Session-Based Recommendation
LLMGR uses an LLM as the semantic engine for session-based recommendation and a GNN as the ranker. Covers the hybrid encoding layer, two-stage prompt tuning, ~8.68% HR@20 lift, and how to deploy without running an LLM …
Graph Neural Networks for Learning Equivariant Representations of Neural Networks
Represent a neural network as a directed graph (neurons as nodes, weights as edges) and use a GNN to produce permutation-equivariant embeddings. The right symmetry unlocks generalisation prediction, network …