Despite its great success, the Standard Model of particle physics predicts the Higgs mass to be unstable at quantum level, which is known as naturalness problem. Various theories to address this problem have been proposed in last several decades, e.g. supersymmetry. In this thesis, we explore several aspects of this topic, ranging from theory study to collider analysis and to the design of data analysis algorithm. We first explore an ultraviolet extension of Twin Higgs model in which the radial mode of global symmetry breaking is itself a pseudo-Nambu-Goldstone Boson. This “turtle” structure raises the scale of new colored states, in exchange for additional states in the Higgs sector, making multiple scalars the definitive signature of naturalness in this context. We then introd...[ Read more ]

Despite its great success, the Standard Model of particle physics predicts the Higgs mass to be unstable at quantum level, which is known as naturalness problem. Various theories to address this problem have been proposed in last several decades, e.g. supersymmetry. In this thesis, we explore several aspects of this topic, ranging from theory study to collider analysis and to the design of data analysis algorithm. We first explore an ultraviolet extension of Twin Higgs model in which the radial mode of global symmetry breaking is itself a pseudo-Nambu-Goldstone Boson. This “turtle” structure raises the scale of new colored states, in exchange for additional states in the Higgs sector, making multiple scalars the definitive signature of naturalness in this context. We then introduce new channels to search for extended Higgs sector which is a landmark of physics beyond the SM including theories of naturalness, at the LHC and next-generation pp-colider. Facilitated with the Boost-Decision-Tree method, we show that LHC has a potential to probe a Higgs sector up to O(1) TeV and a future 100TeV pp-collider close to O(10) TeV in the benchmark of the Minimal Supersymmetric Standard Model. We further explore the application of unsupervised learning method, known as novelty detection, to data analysis in particle physics. This method is model-independent and hence is complementary to supervised learning. With a set of density-based novelty evaluators developed in this study and a deep neural network based on autoencoder, we demonstrate the potential capability of novelty detection in exploring new physics at colliders.