Topical seminar: "Transformers for particle collider analyses"

por Ahmed Hammad (KEK)

viernes 13 feb. 2026 12:00 → 13:00 Europe/Madrid

1001-Primera-1-1-1 - Paterna. Seminario (Universe)

Transformer networks were first developed for language translation and have recently shown great success, especially with the rise of the GPT family. In this talk, I will discuss how Transformers can be applied to data analysis in particle collider experiments. The key idea behind Transformers is the attention mechanism; I will explain different types of attention, including self-attention, cross-attention and sparse attention. Finally, I will discuss recent progress in multi-agent AI approaches that combine pretrained large language models with particle physics tasks. 

References: arXiv:2401.00452 , arXiv:2505.03258, arXiv:2601.21015