Extreme ultraviolet lithography (EUVL) patterned mask defect detection is a major issue that must be addressed to realize EUVL-based device fabrication. We have designed projection electron microscope (PEM) optics for integration into a mask inspection system, and the resulting PEM system performs well in half-pitch (hp) 16-nm-node EUVL patterned mask inspection applications. A learning system has been used in this PEM patterned mask inspection tool. The PEM identifies defects using the “defectivity” parameter that is derived from the acquired image characteristics. The learning system has been developed to reduce the labor and the costs associated with adjustment of the PEM’s detection capabilities to cope with newly defined mask defects. The concepts behind this learning system and the parameter optimization flow are presented here. The learning system for the PEM is based on a library of registered defects. The learning system then optimizes the detection capability by reconciling previously registered defects with newly registered defects. Functional verification of the learning system is also described, and the system’s detection capability is demonstrated by applying it to the inspection of hp 11-nm EUV masks. We can thus provide a user-friendly mask inspection system with reduced cost of ownership.