AI 기반 무기체계의 성능평가를 위한 최적 척도 결정 방법 연구

• 영문명: Suggestion of an Optimal Metric Determination Methodology for the Performance Evaluation of AI Weapon Systems
• 발행기관: 한국IT서비스학회
• 저자명: 김민석(Min Seok Kim) 정재웅(Jae-Woong Jung)
• 간행물 정보: 『한국IT서비스학회지』제24권 제3호, 69~79쪽, 전체 11쪽
• 주제분류: 경제경영 > 경영학
• 파일형태: PDF
• 발행일자: 2025.06.30

국문 초록

As approved in Russia-Ukraine war and Israel-Hamas war, artificial intelligence (AI) technology will determine the aspect of warfare, and thus, many countries including Republic of Korea are doing their best to develop various types of AI weapon systems. To apply the AI weapon system in warfare, it is a prerequisite to appropriately evaluate the performance of the AI weapon system using optimal performance evaluation metric. There are many types of metrics for the performance evaluation of AI weapon systems (e.g., accuracy, precision, F1 score), and each metric has its own purpose, advantages and disadvantages. Therefore, it is of significant importance to select the optimal performance evaluation metric that fits to the characteristics of a specific AI weapon system so that overestimation or underestimation of the actual performance of the AI weapon system can be avoided. Because classification and regression AI models are frequently applied for the development of the AI weapon system, the purpose of this paper is to suggest the methodology (i.e., flowchart) that can choose the optimal metric for the performance evaluation of AI weapon systems developed using both models. Each flowchart consists of sequential questions to determine the optimal metric. We believe that the flowcharts suggested in this paper will serve as a standardized guide for the determination of optimal metric for the performance evaluation of AI weapon systems.

목차

1. 서론

2. 연구 절차

3. AI 무기체계의 성능평가에 적용 가능한 척도 범위 설정

4. AI 기반 무기체계 성능평가 척도 결정 시 고려사항 도출

5. AI 기반 무기체계 최적 척도 결정 방법론 제안

6. 결론

참고문헌

키워드

Artificial Intelligence (AI) AI Weapon System Performance Evaluation Metric Classification Model Regression Model