• KEPCO Journal on Electric Power and Energy
• /
• 제6권4호
• /
• pp.399-404
• /
• 2020

There are many attempts to apply AI technology to diagnose facilities or improve the work efficiency of the power industry. The emergence of new machine learning technologies, such as deep learning, is accelerating the digital transformation of the power sector. The problem is that traditional power systems face security risks when adopting state-of-the-art AI systems. This adoption has convergence characteristics and reveals new cybersecurity threats and vulnerabilities to the power system. This paper deals with the security measures and implementations of the power system using machine learning. Through building a commercial facility operations forecasting system using machine learning technology utilizing power big data, this paper identifies and addresses security vulnerabilities that must compensated to protect customer information and power system safety. Furthermore, it provides security guidelines by generalizing security measures to be considered when applying AI.

• Journal of information and communication convergence engineering
• /
• 제21권1호
• /
• pp.24-31
• /
• 2023

The solid-state drive (SSD) possesses higher input and output speeds, more resistance to physical shock, and lower latency compared with regular hard disks; hence, it is an increasingly popular storage device. However, tiny components on an internal printed circuit board (PCB) hinder the manual detection of malfunctioning components. With the rapid development of artificial intelligence technologies, automatic detection of components through convolutional neural networks (CNN) can provide a sound solution for this area. This study proposes applying the YOLOv5 model to SSD PCB component detection, which is the first step in detecting defective components. It achieves pioneering state-of-the-art results on the SSD PCB dataset. Contrast experiments are conducted with YOLOX, a neck-and-neck model with YOLOv5; evidently, YOLOv5 obtains an mAP@0.5 of 99.0%, essentially outperforming YOLOX. These experiments prove that the YOLOv5 model is effective for tiny object detection and can be used to study the second step of detecting defective components in the future.

• 전자통신동향분석
• /
• 제38권5호
• /
• pp.12-22
• /
• 2023

The development of neural networks is evolving towards the adoption of transformer structures with attention modules. Hence, active research focused on extending the concept of lightweight neural network algorithms and hardware acceleration is being conducted for the transition from conventional convolutional neural networks to transformer-based networks. We present a survey of state-of-the-art research on lightweight neural network algorithms and hardware architectures to reduce memory usage and accelerate both inference and training. To describe the corresponding trends, we review recent studies on token pruning, quantization, and architecture tuning for the vision transformer. In addition, we present a hardware architecture that incorporates lightweight algorithms into artificial intelligence processors to accelerate processing.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 춘계학술대회
• /
• pp.47-49
• /
• 2022

Robot navigation has seen a major improvement since the the rediscovery of the potential of Artificial Intelligence (AI) and the attention it has garnered in research circles. A notable achievement in the area was Deep Learning (DL) application in computer vision with outstanding daily life applications such as face-recognition, object detection, and more. However, robotics in general still depend on human inputs in certain areas such as localization, navigation, etc. In this paper, we propose a study case of robot navigation based on deep reinforcement technology. We look into the benefits of switching from traditional ROS-based navigation algorithms towards machine learning approaches and methods. We describe the state-of-the-art technology by introducing the concepts of Reinforcement Learning (RL), Deep Learning (DL) and DRL before before focusing on visual navigation based on DRL. The case study preludes further real life deployment in which mobile navigational agent learns to navigate unbeknownst areas.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2020년도 추계학술발표대회
• /
• pp.168-171
• /
• 2020

초저지연성을 요구하는 5G 네트워크 환경에서 기기의 핸드오버를 능동적으로 조절하는 시스템의 중요성이 대두되고 있으며, 특히 핸드오버 시 기기의 이동성을 예측하는 것은 필수적이다. 딥러닝 모델의 일종인 생성적 적대 신경망은 두 신경망 사이의 경쟁 구도를 이용하여 두 신경망의 성능을 모두 높이는 목적으로 사용된다. 본 논문에서는 주로 데이터 생성 모델로 사용되는 생성적 적대 신경망을 이용하여 무선 네트워크 환경에서 기기의 이동성을 예측하는 시스템을 개발하였다. 이를 통해 실제 모바일 네트워크 환경에 적용되었을 경우 핸드오버 속도를 높이도록 한다.

• 한국컴퓨터정보학회:학술대회논문집
• /
• 한국컴퓨터정보학회 2023년도 제67차 동계학술대회논문집 31권1호
• /
• pp.85-88
• /
• 2023

미국 콜로라도주 박람회 미술전에서 신인 디지털 아티스트 부문에서 1위를 차지한 게임 디자이너인 제이슨 앨런의 작품 스페이스오페라 극장'이 AI Image generator Midjourney를 활용해서 완성된 작품이라는 것이 알려지면서 창작과 예술 분야에 AI 활용이라는 논쟁이 가속화되고 있다. 창작과 예술을 돕는 탁월한 기능을 가진 툴로 바라보거나 창작과 예술 활동에 아이디어를 제공하고 작품을 구체화하는 과정의 조력자로 환영하는 입장과 예술가의 작품을 허가 없이 훔쳐서 만들어 낸 이미지일 뿐이라는 이상도 이하도 아니며 도덕적으로 허락되어서는 안되다는 입장이 크게 충돌하고 있다. 하루가 다르게 빠르게 발전하고 있는 주요 AI Image generator를 살펴보고 창작과 예술 분야에 AI 활용은 어떤 변화를 가져올지, AI 활용의 긍정적인 측면을 예측하고 연구해 보고자 한다.

• 한국컴퓨터정보학회:학술대회논문집
• /
• 한국컴퓨터정보학회 2023년도 제68차 하계학술대회논문집 31권2호
• /
• pp.529-532
• /
• 2023

본 논문은 소셜 네트워크 이용자의 텍스트 데이터를 대상으로, 트랜스포머 계열의 언어모델을 전이학습해 이용자의 MBTI 성격 유형을 분류한 국내 첫 연구이다. Kaggle MBTI Dataset을 대상으로 RoBERTa Distill, DeBERTa-V3 등의 사전 학습모델로 전이학습을 해, MBTI E/I, N/S, T/F, J/P 네 유형에 대한 분류의 평균 정확도는 87.9181, 평균 F-1 Score는 87.58를 도출했다. 해외 연구의 State-of-the-art보다 네 유형에 대한 F1-Score 표준편차를 50.1% 낮춰, 유형별 더 고른 분류 성과를 보였다. 또, Twitter, Reddit과 같은 글로벌 소셜 네트워크 서비스의 텍스트 데이터를 추가로 분류, 트랜스포머 기반의 MBTI 분류 방법론을 확장했다.

• 한국항공운항학회지
• /
• 제31권3호
• /
• pp.172-177
• /
• 2023

The main purpose of the research was to review the global trends of airport's smart security technologies. Moreover, using the case studies of airport using smart security, this paper tried to propose the implication how the findings through the case studies may be important for airport policy and will impact the future research of airport operation. It is expected in the future the aviation security technology with biometric information evolves from single identification to multiple identification technology which has combined application of iris, vein and others. Facing post COVID-19 era, the number of passengers traveling through airports continues increase dramatically and the risks as well, the role of AI becomes even more crucial. With AI based automated security robotics airport operators could effectively handle the growing passenger and cargo volume and address the associated issues Smart CCTV analysis with A.I. and IoT applying solutions could also provide significant support for airport security.

• ETRI Journal
• /
• 제45권6호
• /
• pp.929-938
• /
• 2023

The detection of all the symbols transmitted simultaneously in multiuser systems using limited wireless resources is challenging. Traditional model-based methods show high performance with perfect channel state information (CSI); however, severe performance degradation will occur if perfect CSI cannot be acquired. In contrast, data-driven methods perform slightly worse than model-based methods in terms of symbol error ratio performance in perfect CSI states; however, they are also able to overcome extreme performance degradation in imperfect CSI states. This study proposes a novel deep learning-based method by improving a state-of-the-art data-driven technique called deep soft interference cancellation (DSIC). The enhanced DSIC (EDSIC) method detects multiuser symbols in a fully sequential manner and uses an efficient neural network structure to ensure high performance. Additionally, error-propagation mitigation techniques are used to ensure robustness against channel uncertainty. The EDSIC guarantees a performance that is very close to the optimal performance of the existing model-based methods in perfect CSI environments and the best performance in imperfect CSI environments.

• Journal of Information Technology Applications and Management
• /
• 제30권4호
• /
• pp.11-28
• /
• 2023

This paper proposes a method of integrating ChatGPT with traditional chatbot systems to enhance conversational artificial intelligence(AI) and create more efficient conversational systems. Traditional chatbot systems are primarily based on classification models and are limited to intent classification and simple response generation. In contrast, ChatGPT is a state-of-the-art AI technology for natural language generation, which can generate more natural and fluent conversations. In this paper, we analyze the business service areas that can be integrated with ChatGPT and traditional chatbots, and present methods for conducting conversational scenarios through case studies of service types. Additionally, we suggest ways to integrate ChatGPT with traditional chatbot systems for intent recognition, conversation flow control, and response generation. We provide a practical implementation example of how to integrate ChatGPT with traditional chatbots, making it easier to understand and build integration methods and actively utilize ChatGPT with existing chatbots.