[2024 Spring] Graph Machine Learning - Part 5: Introduction to Graph Attention Network

Описание к видео [2024 Spring] Graph Machine Learning - Part 5: Introduction to Graph Attention Network

Graph Attention Networks (GATs), a key advancement from Graph Convolutional Networks (GCNs) (find out more on our part 4 video!).

Unlike GCNs, which use static normalization coefficients, GATs employ dynamic weighting factors derived through a mechanism known as self-attention—a core feature in some of the most influential deep learning models, such as transformers, BERT, and GPT-3.

Learn more on the topic:
- GitHub Wiki on the topic: https://github.com/ua-datalab/GraphML
- UA Data Lab Machine Learning Repositories: https://github.com/ua-datalab/MLWorks...
- Data Science Institute Calendar: https://datascience.arizona.edu/calendar

Комментарии

Информация по комментариям в разработке

Похожие видео