New Combinatorial Residual Network Classifier Developed for Particle Physics

Researchers have developed a new classifier called the Combinatorial Residual Network (CoReNet) designed to handle complex combinatorial problems in particle physics, specifically for identifying the decay chains of heavy particles. This network architecture encodes the assumed decay hierarchy, allowing it to evaluate all jet assignment hypotheses with shared weights and combine them using learned weights. CoReNet demonstrates strong performance on simulated Higgs boson pair production and $HH o bar{b}W^+W^-$ events, outperforming larger transformer-based networks while requiring significantly fewer parameters and training time. AI

IMPACT Introduces a novel neural network architecture for complex combinatorial problems in physics, potentially improving efficiency and accuracy in particle identification.

RANK_REASON The cluster contains a research paper detailing a new model architecture for particle physics. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv cs.LG →

• arXiv
• Combinatorial Residual Network
• Higgs boson
• Hugging Face
• QCD
• Tirso Alejandro Gómez Espinosa
• transformer

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →