关键词: 前馈式神经元网络, 丢失能量, 唯象性研究, 质量重建

Abstract: The main difficulty of heavy Higgs search through H°→ZZ (WW)→llvv channel at LHC pp collider comes from the large energy loss, so it is impossible to reconstruct its invariant mass by conventional method, its existence can only be seen through Jacobi peak of p_T^Z or m_T distribution. As a phenomenological study based on Monte Carlo simulation, a feed-forward neural network is designed to reconstruct Higgs invariant mass for giving correct M_H mass position and satisfactory width, as well as the background rejection ability. This method is proved to be quite suitable for new praticle search and mass measurement in experiment.

Key words: feed-forward neural network, energy loss, phenomenological study, mass reconstruction