Graph Representational Learning Book

Livro digital

Título:
Graph Representational Learning Book

Autor:
William L. Hamilton

Categoria:
Tecnologia > IA

Doador:
Raffaello D. N.

Sinopse:
Graph Representation Learning tackles the hard problem of learning from data where relationships matter as much as individual items, and the table of contents makes that progression explicit: from graph basics and network analysis, to node embeddings, multi-relational knowledge graphs, and then the full graph neural network stack. William L. Hamilton structures the book as a practical survey of the field’s main ideas and their evolution. It opens with graph statistics, kernels, overlap measures, and spectral methods, then moves into encoder-decoder views of embeddings, random-walk methods, and knowledge graph reconstruction. The later sections cover message passing, neighborhood aggregation, attention, pooling, practical training concerns, and the theoretical links to convolutions, probabilistic graphical models, and graph isomorphism. The result is a coherent map of graph machine learning rather than a narrow method guide. Readers get both the conceptual foundations and the modern deep-learning toolkit, with enough breadth to understand where graph representation learning came from and enough depth to apply GNN-based methods to real relational data.