29 October 2025 to 2 November 2025
河南省新乡市 (Xinxiang, Henan)
Asia/Shanghai timezone

Machine learning-enhanced search for 95 GeV light Higgs in N2HDM at CEPC

31 Oct 2025, 17:20
20m
平原AB厅

平原AB厅

Theory Parallel 4

Speaker

亚博 董 (河南大学)

Description

The Flipped Next-to-2HDM (N2HDM-F) provides an elegant solution to the possible 95 GeV signal excesses, and the proposed Future Circular Electron-Positron Collider (CEPC) offers a promising opportunity to probe this light Higgs boson ($S$). A Monte Carlo (MC) simulation of the $e^+e^-\to Z(\to\mu^+\mu^-)S(\to \tau^+\tau^-)$ process is performed to study the discovery potential for $S$ and the measurement precision of its $SZZ$ coupling and $S\to\tau^+\tau^-$ branching ratio. The deep neural network architecture of the More-Interaction Particle Transformer (MIParT) is employed to enhance both the discovery potential and the measurement precision. Through comprehensive feature extraction of all final-state particles, more than 100\% improvement is achieved in both sensitivity and precision. Furthermore, by combining with the $b\bar{b}$ decay channel, more than 85\% of the surviving N2HDM-F parameter samples can be probed at the $5\sigma$ level with an integrated luminosity of $20\mathrm{ab}^{-1}$ at the CEPC.

Primary authors

亚博 董 (河南大学) Dr 坤 王 (University of Shanghai for Science and Technology) Prof. Haijun Yang (Shanghai Jiao Tong University) Prof. 经亚 朱 (河南大学)

Presentation materials