• Journal of the Microelectronics and Packaging Society
• /
• v.30 no.1
• /
• pp.30-41
• /
• 2023

With the aggressive down-scaling of advanced integrated circuits (ICs), electronic packages have become the bottleneck of both reliability and performance of whole electronic systems. In order to resolve the reliability issues, Institute of Electrical and Electronics Engineers (IEEE) laid down a roadmap on fault detection and diagnosis (FDD), thrusting the digital twin: a combination of reliability physics and artificial intelligence (AI). In this paper, we especially review research works regarding the signal-based FDD approaches on the electronic packages. We also discuss the research trend of FDD utilizing AI techniques.

• Advanced Industrial SCIence
• /
• v.1 no.1
• /
• pp.28-34
• /
• 2022

Many studies have been conducted on software quality evaluation. Recently, as artificial intelligence-related software has been developed a lot, research on methods for evaluating artificial intelligence functions in existing software is being conducted. Software evaluation has been based on eight quality characteristics: functional suitability, reliability, usability, maintainability, performance efficiency, portability, compatibility, and security. Research on the part that needs to be confirmed through evaluation of the function of the intelligence part is in progress. This study intends to introduce the contents of the evaluation method in this part. We are going to propose a quality evaluation method for artificial intelligence software by presenting the existing software quality evaluation method and the part to be considered in the AI part.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.01a
• /
• pp.217-218
• /
• 2023

인공지능은 디지털 대전환 시대의 핵심적인 기술로 사회 전반에 변화를 주도하였다. 우리나라는 인공지능을 이해하고 이를 활용하는 역량을 길러주기 위해 전 국민을 대상으로 교육을 진행하고 있다. 그러나 초등학생 대상 인공지능 교육 프로그램은 체험 및 놀이 실습으로 한정되어 교육적 효과에 한계가 있다. 그러므로 본 연구에서는 생산적 실패를 활용하여 인공지능에 대한 개념적 이해 및 실생활 전이를 촉진하는 교육 프로그램을 개발하였다. 연구 대상은 초등학교 5~6학년이며 2022 개정 교육과정에서 강조하는 자기 주도적 학습 역량과 실생활 연계 교육을 반영하여 설계한 6차시 분량의 프로그램이다. 본 연구에서 개발한 교육 프로그램은 향후 타당성 및 신뢰도 검증을 거쳐 현장에 적용하는 후속 연구로 이어질 것이다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.520-522
• /
• 2022

의료 인공지능은 특정 진단에서 높은 정확도를 보이지만 모델의 신뢰성 문제로 인해 활발하게 쓰이지 못하고 있다. 이에 따라 인공지능 모델의 진단에 대한 원인 설명의 필요성이 대두되었고 설명가능한 의료 인공지능에 관한 연구가 활발히 진행되고 있다. 하지만 MRI 등 의료 영상 인공지능 분야에서 주로 진행되고 있으며, 이미지 형태가 아닌 전자의무기록 데이터 (Electronic Health Record, EHR) 를 기반으로 한 모델의 설명가능성 연구는 EHR 데이터 자체의 복잡성 때문에 활발하게 진행 되지 않고 있다. 본 논문에서는 전자의무기록 데이터인 MIMIC-III (Medical Information Mart for Intensive Care) 를 전처리 및 그래프로 표현하고, GCT (Graph Convolutional Transformer) 모델을 학습시켰다. 학습 후, 어텐션 흐름 그래프를 시각화해서 모델의 예측에 대한 직관적인 설명을 제공한다.

• Journal of Internet Computing and Services
• /
• v.23 no.3
• /
• pp.111-123
• /
• 2022

Artificial intelligence, which is developing as the core of an intelligent information society, is bringing convenience and positive life changes to humans. However, with the development of artificial intelligence, human rights and property are threatened, and ethical problems are increasing, so alternatives are needed accordingly. In this study, the most controversial artificial intelligence ethics problem in the dysfunction of artificial intelligence was aimed at researching and developing artificial intelligence ethical measurement indicators to protect human personality rights and property first under artificial intelligence ethical principles and components. In order to research and develop artificial intelligence ethics measurement indicators, various related literature, focus group interview(FGI), and Delphi surveys were conducted to derive 43 items of ethics measurement indicators. By survey and statistical analysis, 40 items of artificial intelligence ethics measurement indicators were confirmed and proposed through descriptive statistics analysis, reliability analysis, and correlation analysis for ethical measurement indicators. The proposed artificial intelligence ethics measurement indicators can be used for artificial intelligence design, development, education, authentication, operation, and standardization, and can contribute to the development of safe and reliable artificial intelligence.

• Journal of Science Education
• /
• v.45 no.1
• /
• pp.105-117
• /
• 2021

The purpose of this study is to analyze the gender differences of elementary, junior high, and high school students in the artificial intelligence ethics awareness (hereinafter referred to as AIEA). This is a study to investigate whether there is a gender difference in the AIEA, and if so, when the gender difference will occur. This study was conducted with 198 elementary school students (98 female students, 100 male students), 265 middle school students (166 female students, 99 male students), and 114 high school students (58 female students and 56 male students) in I Metropolitan City. The results are as follows: First, a gender difference in the AIEA between all boys and girls was confirmed. Second, the gender difference in the AIEA tended to be solidified as the school age increased from elementary school to middle school and high school. Third, female students at all stages of elementary school, junior high school, and high school are not yet very reliable in artificial intelligence, and there is a greater concern about non-discrimination than boys. It turns out that they have a negative position on permission to enter the territory. Fourth, the interaction effects of school age and gender have been identified in 'stability and reliability,' and in 'permit and limit' categories. Taken together, these results show that an educational strategy that approaches the gender equality perspective of the educational program is necessary so that there will be no gender difference in the AIEA during artificial intelligence education activities.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.5
• /
• pp.147-157
• /
• 2022

The purpose of this study is to establish policy recommendations for the promotion of AI service impact assessment based on the definition of impact assessment and analysis of domestic and foreign AI service impact assessment cases. The direction of implementation was analyzed based on the case of impact evaluation promoted in various fields at home and abroad and the case of impact evaluation at home and abroad of artificial intelligence services. As a step-by-step implementation plan, in the first stage, quantitative indicators such as AI level survey-based economic effects are developed, and in the second stage, information culture such as safety and reliability and artificial intelligence ethics described in the Framework Act on Intelligence Information, social, economic, information protection, and people's daily lives are prepared. In the third stage, discussion on detailed metrics and methods will be expanded and impact assessment results will be evaluated. This study requires analysis through various participants such as policy designers, artificial intelligence service developers, and civic groups in the future.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2022.05a
• /
• pp.131-131
• /
• 2022

기후변화의 영향으로 극치강우의 변동성이 커지고 있으며 계획빈도를 초과하는 폭우로 피해가 증가하고 있다. 기존의 물리기반의 홍수예측모델은 개념적 및 구조적 제약과 함께 다양한 유역조건 및 수문기상 조건에 기인한 강우-유출 관계의 불확실성을 고려하는 데 한계가 있다. 특히 한정된 홍수 사상을 통해 구축된 관측 자료로 인해 새로운 홍수 사상 예측 능력이 저조할 수밖에 없다. 따라서 기존 물리모형 기반의 홍수예측과 함께, 딥러닝(deep learning) 모형을 고려한 홍수예측 모델 개발과 개선이 필요하다. 본 연구에서는 다양한 분야에서 활용되는 인공지능(artificial intelligence, AI) 기술을 종합적으로 검토하고, 홍수 예측 측면에서의 활용 가능성 및 신뢰성을 고려하여 AI 기법을 채택하였다. 한강수계에 존재하는 댐 중 일부를 선정하여 대상 댐의 수문·기상학적 자료를 전처리한 후, 인공지능 기반의 홍수예측모형을 구축 및 최적화하였다. 다양한 예측인자와 모델 구성으로 홍수예측력에 대한 평가를 다각적으로 수행함으로써 홍수예측모델의 신뢰성을 제고하였다. 전반적으로 우수한 결과를 도출하였고, 유역면적이 작을수록 결과가 좋았다. 이는 넓은 유역일수록 복잡한 강우-유출 과정이 내재되어 있기 때문으로 판단되며, 넓은 유역에는 본 연구에서 활용한 자료에 추가적인 자료를 도입하여 모형 개선이 이루어져야 할 것으로 판단하였다. 수문 예측 연구에 통계모형이나 기계학습모형의 적용은 많이 있었지만, 딥러닝 기법 활용은 새로운 시도라는 점에서 의미가 있다.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.115-123
• /
• 2024

The purpose of this study is to contribute to how the advantages of artificial intelligence (AI) services and the associated limitations can be simultaneously overcome, using the keywords AI and risk management. To achieve this, two cases were introduced: (1) presenting a risk monitoring process utilizing AI and (2) introducing an operational toolkit to minimize the emerging limitations in the development and operation of AI services. Through case analysis, the following implications are proposed. First, as AI services deeply influence our lives, the process are needed to minimize the emerging limitations. Second, for effective risk management monitoring using AI, priority should be given to obtaining suitable and reliable data. Third, to overcome the limitations arising in the development and operation of AI services, the application of a risk management process at each stage of the workflow, requiring continuous monitoring, is essential. This study is a research effort on approaches to minimize limitations provided by advancing artificial intelligence (AI). It can contribute to research on risk management in the future growth and development of the related market, examining ways to mitigate limitations posed by evolving AI technologies.

• Convergence Security Journal
• /
• v.22 no.3
• /
• pp.101-110
• /
• 2022

As the cases of using artificial intelligence in various fields increase, attempts to solve various issues through artificial intelligence in the intrusion detection field are also increasing. However, the black box basis, which cannot explain or trace the reasons for the predicted results through machine learning, presents difficulties for security professionals who must use it. To solve this problem, research on explainable AI(XAI), which helps interpret and understand decisions in machine learning, is increasing in various fields. Therefore, in this paper, we propose an explanatory AI to enhance the reliability of machine learning-based intrusion detection prediction results. First, the intrusion detection model is implemented through XGBoost, and the description of the model is implemented using SHAP. And it provides reliability for security experts to make decisions by comparing and analyzing the existing feature importance and the results using SHAP. For this experiment, PKDD2007 dataset was used, and the association between existing feature importance and SHAP Value was analyzed, and it was verified that SHAP-based explainable AI was valid to give security experts the reliability of the prediction results of intrusion detection models.