Computational electromagnetism plays important roles both to design the electric facilities and to assess the influence of electromagnetic fields; for example, transformer, motor, burn wound, and hyperthermic potentiation. However, as the ability of computers progresses and the demand of more precise approximation, the number of Degrees Of Freedom (DOF) of computational models derived from conventional numerical schemes becomes larger even in case of adaptive mesh refinements. In this mini-symposium, we discuss on the accuracy and efficiency of novel computational methods for electromagnetic field problems from mathematical and engineering points of view. First, we discuss efficient numerical schemes to compute directly such large-scale computational models within the required computational costs, for example, based on Domain Decomposition Methods (DDM) with parallel computations; see [1]. Second, we discuss new approaches based on machine learning and artificial intelligence. Third, we welcome to discuss novel schemes not mentioned above. Moreover, we also consider other related problems as well as electromagnetic field problems because we encounter a kind of interaction problems with electromagnetic field ones in practical situations; see [2].