AI-ordlista
Den kompletta ordlistan över AI
Graph Convolutional Networks (GCN)
Fundamental GNN architecture applying convolution operations on graphs by aggregating features from direct neighbors.
Graph Attention Networks (GAT)
GNN incorporating attention mechanisms to weight the influence of each neighbor differently during feature aggregation.
GraphSAGE
Inductive aggregation algorithm that samples and aggregates neighbor features to generate node embeddings.
Message Passing Neural Networks
General paradigm where nodes exchange messages iteratively to update their latent representations.
Heterogeneous Graph Neural Networks
GNNs specialized in processing graphs containing multiple types of nodes and/or edges with different relationships.
Temporal Graph Neural Networks
GNN architectures designed to model dynamic graphs whose structure evolves over time.
Graph Autoencoders
Unsupervised GNN models learning compact representations of graphs via reconstruction of the structure or attributes.
Graph Generative Models
GNNs capable of generating new graph structures by learning the distribution of training graphs.
Spatial vs Spectral GNNs
Two complementary approaches: spatial defines convolution directly on neighbors, spectral uses graph theory via Fourier transform.
Graph Transformers
Hybridization between GNN and Transformer architectures, combining message passing and global attention mechanisms.
Knowledge Graph Embedding
GNN techniques specialized in learning vector representations for structured knowledge graphs.
Molecular Graph Neural Networks
GNN optimized for molecular property prediction and drug discovery by treating molecules as graphs.