• 한국인터넷방송통신학회논문지
• /
• 제22권5호
• /
• pp.147-157
• /
• 2022

본 연구의 목적은 영향평가에 대한 정의부터 국내외 인공지능 서비스 영향평가 사례에 대한 분석을 바탕으로 인공지능 서비스 영향평가 추진에 대한 정책적 제언을 수립하는 것이다. 이를 위해서 국내외 타 분야에서 추진된 영향평가 사례, 인공지능 서비스 국내외 영향평가 사례를 바탕으로 추진 방향을 분석하였다. 국내 인공지능 서비스 영향평가는 다소 광범위하고 포괄적이며, 시점도 사전적 예방 수단에 그치지 않고 상시적·사후적 위험성 관리를 예정하고 있다. 단계별 추진 방안으로 1단계에는 AI 수준 조사 기반의 경제적 효과 등의 정량적 지표를 개발한 후, 2단계에서는 지능정보화 기본법에 기술된 안전성 및 신뢰성, 인공지능 윤리 등 정보문화, 고용·노동 등 사회·경제, 정보보호, 국민의 일상생활에 미치는 영향에 미치는 분야별 평가체계를 마련한다. 3단계에서는 세부 측정지표나 방식 등의 논의를 확대하고 영향평가 결과가 인공지능 정책에 반영하는 환류 체계 포함된다면 국내의 인공지능 경쟁력 강화에 도움이 되는 정책 수단이 될 수 있다는 것을 제시하였다. 본 연구는 향후 정책 설계자, 인공지능 서비스 개발자, 시민단체 등 다양한 참여자를 통한 분석이 요구된다.

• 한국컴퓨터정보학회논문지
• /
• 제26권1호
• /
• pp.275-281
• /
• 2021

본 연구의 목적은 인공지능예술작품 구매할 때 소비자들이 어떤 감지가 있는지, 그리고 구매의 향과 어떤 관계가 있는지 살펴보는데 있다. 본 연구에서 고객감지가치가 제품감지가치, 서비스감지가치 그리고 사회감지가치 총 3가지를 제시하였다. 이를 바탕으로 고객감지가치와 구매의향 간의 모델을 구축하였다. 연구를 위해 데이터 수집은 온라인 설문 조사를 실시하였다. SPSS24.0와 AMOS24.0을 통해 수집한 데이터의 신뢰성, 타당성 및 구조 방정식 분석을 통해 가설 검증을 하였다. 검정결과를 보면 제품인지가치와 서비스인지가치는 소비자의 온라인 구매의향에 긍정적인 영향을 미친다. 그러나 사회인지가치가 소비자의 구매의향에 영향을 주지 않는 결과가 나타났다.

• ETRI Journal
• /
• 제43권4호
• /
• pp.630-639
• /
• 2021

Autonomous driving requires a computerized perception of the environment for safety and machine-learning evaluation. Recognizing semantic information is difficult, as the objective is to instantly recognize and distinguish items in the environment. Training a model with real-time semantic capability and high reliability requires extensive and specialized datasets. However, generalized datasets are unavailable and are typically difficult to construct for specific tasks. Hence, a light detection and ranging semantic dataset suitable for semantic simultaneous localization and mapping and specialized for autonomous driving is proposed. This dataset is provided in a form that can be easily used by users familiar with existing two-dimensional image datasets, and it contains various weather and light conditions collected from a complex and diverse practical setting. An incremental and suggestive annotation routine is proposed to improve annotation efficiency. A model is trained to simultaneously predict segmentation labels and suggest class-representative frames. Experimental results demonstrate that the proposed algorithm yields a more efficient dataset than uniformly sampled datasets.

• 대한의용생체공학회:의공학회지
• /
• 제42권5호
• /
• pp.241-249
• /
• 2021

Recently, artificial intelligence for diagnosis system of obstetric diseases have been actively studied. Artificial intelligence diagnostic assist systems, which support medical diagnosis benefits of efficiency and accuracy, may experience problems of poor learning accuracy and reliability when inappropriate images are the model's input data. For this reason, before learning, We proposed an algorithm to exclude unread cervical imaging. 2,000 images of read cervical imaging and 257 images of unread cervical imaging were used for this study. Experiments were conducted based on the statistical method Radiomics to extract feature values of the entire images for classification of unread images from the entire images and to obtain a range of read threshold values. The degree to which brightness, blur, and cervical regions were photographed adequately in the image was determined as classification indicators. We compared the classification performance by learning read cervical imaging classified by the algorithm proposed in this paper and unread cervical imaging for deep learning classification model. We evaluate the classification accuracy for unread Cervical imaging of the algorithm by comparing the performance. Images for the algorithm showed higher accuracy of 91.6% on average. It is expected that the algorithm proposed in this paper will improve reliability by effectively excluding unread cervical imaging and ultimately reducing errors in artificial intelligence diagnosis.

• 산경연구논집
• /
• 제11권6호
• /
• pp.41-53
• /
• 2020

Purpose: The purpose of this study is to review the previous studies on the characteristics of the image search service provided by using artificial intelligence, the social impact characteristics, and the moderating effect of perceived anthropomorphism, and conduct empirical analysis to identify the constituent factors affecting purchase intention. To clarify. Through this, I tried to present theoretical and practical implications. Research design, data, and methodology: Research design was that characteristics of image search service (ubiquity and information quality) and social impact characteristics (subjective norms, electronic word of mouth marketing) are affected by mediation of satisfaction and flow, therefore, control of perceived anthropomorphism have an effect on purchase intention to increase. For analysis, research conducted literature review, and developed questionnaires, so that EM firm which is a specialized research institute has collected data. This was conducted on 410 people between the 20s and 50s who have mobile shopping experiences. SPSS Statistics 23 and AMOS 23 had been used to perform necessary analysis such as exploratory factor analysis, reliability analysis, feasibility analysis, and structural equation modeling based on this data. Results: first, ubiquity, information quality and subjective norms were found to have a positive effect on purchase intention through satisfaction and flow parameters. Second, satisfaction and flow were found to have a mediating effect between ubiquity, information quality, and subjective norms and purchase intentions. However, there was no mediating effect between eWOM information and purchase intention. Third, perceived anthropomorphism was found to have a moderating effect between information quality and satisfaction, and it was found that there was no moderating effect on the relationship between information quality and flow. Conclusions: The information quality of image search services using artificial intelligence has a positive effect on satisfaction, and it has been found that there is a positive moderate effect of perceived anthropomorphism in this relationship, which may be an academic contribution to the distribution science utilizing artificial intelligence. Therefore, it is possible to propose a distribution strategy that improves purchase intention by utilizing image search service and anthropomorphism in practical business and providing a more enjoyable immersive experience to customers.

• 대한의용생체공학회:의공학회지
• /
• 제43권6호
• /
• pp.390-398
• /
• 2022

Ultrasound examination using ultrasound equipment is an ultrasound device that images human organs using sound waves and is used in various areas such as diagnosis, follow-up, and treatment of diseases. However, if the quality of ultrasound equipment is not guaranteed, the possibility of misdiagnosis increases, and the diagnosis rate decreases. Accordingly, The Korean Society of Radiology and Korea society of Ultrasound in Medicine presented guidelines for quality management of ultrasound equipment using ATS-539 phantom. The DenseNet201 classification algorithm shows 99.25% accuracy and 5.17% loss in the Dead Zone, 97.52% loss in Axial/Lateral Resolution, 96.98% accuracy and 20.64% loss in Sensitivity, 93.44% accuracy and 22.07% loss in the Gray scale and Dynamic Range. As a result, it is the best and is judged to be an algorithm that can be used for quantitative evaluation. Through this study, it can be seen that if quantitative evaluation using artificial intelligence is conducted in the qualitative evaluation item of ultrasonic equipment, the reliability of ultrasonic equipment can be increased with high accuracy.

• International Journal of Internet, Broadcasting and Communication
• /
• 제15권3호
• /
• pp.191-202
• /
• 2023

This study aims to propose a framework for evaluating the impact of artificial intelligence (AI) services, based on the concept of AI service impact. It also suggests a model for evaluating this impact and identifies relevant factors and measurement approaches for each item of the model. The study classifies the impact of AI services into five categories: ethics, safety and reliability, compliance, user rights, and environmental friendliness. It discusses these five categories from a broad perspective and provides 21 detailed factors for evaluating each category. In terms of ethics, the study introduces three additional factors-accessibility, openness, and fairness-to the ten items initially developed by KISDI. In the safety and reliability category, the study excludes factors such as dependability, policy, compliance, and awareness improvement as they can be better addressed from a technical perspective. The compliance category includes factors such as human rights protection, privacy protection, non-infringement, publicness, accountability, safety, transparency, policy compliance, and explainability.For the user rights category, the study excludes factors such as publicness, data management, policy compliance, awareness improvement, recoverability, openness, and accuracy. The environmental friendliness category encompasses diversity, publicness, dependability, transparency, awareness improvement, recoverability, and openness.This study lays the foundation for further related research and contributes to the establishment of relevant policies by establishing a model for evaluating the impact of AI services. Future research is required to assess the validity of the developed indicators and provide specific evaluation items for practical use, based on expert evaluations.

• 융합보안논문지
• /
• 제23권1호
• /
• pp.27-33
• /
• 2023

Covid-19으로 인한 디지털화는 인공지능 기반의 음성인식 기술을 급속하게 발전시켰다. 그러나 이 기술은 데이터셋이 일부 집단에 편향될 경우 인종 및 성차별과 같은 불공정한 사회적 문제를 초래하고 인공지능 서비스의 신뢰성과 보안성을 열화시키는 요인이 된다. 본 연구에서는 대표적인 인공지능의 CNN(Convolutional Neural Network) 모델인 VGGNet(Visual Geometry Group Network), ResNet(Residual neural Network), MobileNet을 활용한 편향된 데이터 환경에서 정확도에 기반한 불공정성을 비교 및 분석한다. 실험 결과에 따르면 Top1-accuracy에서 ResNet34가 여성과 남성이 91%, 89.9%로 가장 높은 정확도를 보였고, 성별 간 정확도 차는 ResNet18이 1.8%로 가장 작았다. 모델별 성별 간의 정확도 차이는 서비스 이용 시 남녀 간의 서비스 품질에 대한 차이와 불공정한 결과를 야기한다.

• International journal of advanced smart convergence
• /
• 제5권4호
• /
• pp.40-47
• /
• 2016

Journalism robot by using a computer algorithm, while maintaining the precision and reliability of the existing media refers to an article which is automatically created. In this paper, we introduce 'stock robot' of robot journalism which writes securities articles and describe artificial intelligence algorithms in stages. Key steps of stock robot implemented artificial intelligence algorithm through four steps of data collection and storage, key event extraction, article content production, and article production. This research has developed a stock robot that collects and analyzes data on social issues and stock indexes for the last 2 years. In the future, as the algorithm is further developed, it becomes possible to write securities articles quickly and accurately through social issues. It will also provide customized information tailored to the user's preferences.

• 한국정보시스템학회지:정보시스템연구
• /
• 제32권4호
• /
• pp.189-210
• /
• 2023

Purpose This study aims to investigate factors that positively influence the continuous Artificial Intelligence(AI) Learning Continuity of business major students. Design/methodology/approach To evaluate the impact of AI education, a survey was conducted among 119 business-related majors who completed a software/AI course. Frequency analysis was employed to examine the general characteristics of the sample. Furthermore, factor analysis using Varimax rotation was conducted to validate the derived variables from the survey items, and Cronbach's α coefficient was used to measure the reliability of the variables. Findings Positive correlations were observed between business major students' AI Learning Continuity and their AI Interest, AI Awareness, and Data Analysis Capability related to their majors. Additionally, the study identified that AI Project Awareness and AI Literacy Capability play pivotal roles as mediators in fostering AI Learning Continuity. Students who acquired problem-solving skills and related technologies through AI Projects Awareness showed increased motivation for AI Learning Continuity. Lastly, AI Self-Efficacy significantly influences students' AI Learning Continuity.