<p>Ultrafast Electron Diffraction (UED) is an imaging method to study the structure of matter and its dynamics by scattering of a pulsed electron bunch. Its performance depends on electron beam parameters such as beam size and bunch length, which are in turn affected by UED machine settings including solenoid strength and RF phase. However, these beam parameters cannot be optimized simultaneously due to their mutual correlations, such as those arising from space-charge effects. Multi-Objective Bayesian Optimization (MOBO) is a sample-efficient method widely used for the simultaneous optimization of accelerator parameters; however, its computational cost grows rapidly with the dimensionality of the input parameter space. To solve this problem, Multi-Objective Regionalized Bayesian Optimization (MORBO) has been proposed. This method searches the next measurement point only in the trust region whose center and width vary during the optimization process. In this paper, we determined the optimal beam parameter set (Pareto front) at the sample position via MORBO. The parametric space consists of machine settings of Pohang Accelerator Laboratory (PAL) UED facility, which delivers MeV-level electron beams using the infrastructures of the PAL electron Linear Accelerator for Basic Science (PAL-eLABs). Also, to verify whether MORBO is effective in low-dimensional problems, the optimization performance of MORBO was compared with normal MOBO and other conventional optimization methods.</p>