• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.597-599
• /
• 2022

When IoT device performs on-device AI, the device is required to use various AI models selectively according to target service and surrounding environment. Also, AI model can be updated by additional training such as federated learning or adapting the improved technique. Hence, for successful on-device AI, IoT device should acquire various AI models selectively or update previous AI model to new one. In this paper, we propose AI model repository to tackle this issue. The repository supports AI model registration, searching, management, and deployment along with dashboard for practical usage. We implemented it using Node.js and Vue.js to verify it works well.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.24 no.1
• /
• pp.1-8
• /
• 2024

On-device AI technology, which can operate AI models at the edge devices to support real-time processing and privacy enhancement, is attracting attention. As intelligent IoT is applied to various industries, services utilizing the on-device AI technology are increasing significantly. However, general deep learning models require a lot of computational resources for inference and learning. Therefore, various lightweighting methods such as quantization and pruning have been suggested to operate deep learning models in embedded edge devices. Among the lightweighting methods, we analyze how to lightweight and apply deep learning models to edge computing devices, focusing on pruning technology in this paper. In particular, we utilize dynamic and static pruning techniques to evaluate the inference speed, accuracy, and memory usage of a lightweight AI vision model. The content analyzed in this paper can be used for intelligent video control systems or video security systems in autonomous vehicles, where real-time processing are highly required. In addition, it is expected that the content can be used more effectively in various IoT services and industries.

• Electronics and Telecommunications Trends
• /
• v.39 no.4
• /
• pp.82-92
• /
• 2024

This paper introduces the technological development trends in on-device SLMs (Small Language Models). Large Language Models (LLMs) based on the transformer model have gained global attention with the emergence of ChatGPT, providing detailed and sophisticated responses across various knowledge domains, thereby increasing their impact across society. While major global tech companies are continuously announcing new LLMs or enhancing their capabilities, the development of SLMs, which are lightweight versions of LLMs, is intensely progressing. SLMs have the advantage of being able to run as on-device AI on smartphones or edge devices with limited memory and computing resources, enabling their application in various fields from a commercialization perspective. This paper examines the technical features for developing SLMs, lightweight technologies, semiconductor technology development trends for on-device AI, and potential applications across various industries.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.22 no.5
• /
• pp.17-22
• /
• 2022

In this paper, we design a lightweight embedded device that can support intelligent edge computing, and show that the device quickly detects an object in an image input from a camera device in real time. The proposed system can be applied to environments without pre-installed infrastructure, such as an intelligent video control system for industrial sites or military areas, or video security systems mounted on autonomous vehicles such as drones. The On-Device AI(Artificial intelligence) technology is increasingly required for the widespread application of intelligent vision recognition systems. Computing offloading from an image data acquisition device to a nearby edge device enables fast service with less network and system resources than AI services performed in the cloud. In addition, it is expected to be safely applied to various industries as it can reduce the attack surface vulnerable to various hacking attacks and minimize the disclosure of sensitive data.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.51-54
• /
• 2024

본 논문에서는 온 디바이스 국방 AI를 위한 효율적인 학습 방법을 제안한다. 제안하는 방법은 모델 전체를 재학습하는 대신 필요한 부분만 세밀하게 조정하여 계산 비용과 시간을 대폭 줄이는 PEFT 기법의 LoRa를 적용하였다. LoRa는 기존의 신경망 가중치를 직접 수정하지 않고 추가적인 낮은 랭크의 매트릭스를 학습하는 방식으로 기존 모델의 구조를 크게 변경하지 않으면서도, 효율적으로 새로운 작업에 적응할 수 있다. 또한 학습 파라미터 및 연산 입출력에 데이터에 대하여 32비트의 부동소수점(FP32) 대신 부동소수점(FP16, FP8) 또는 정수형(INT8)을 활용하는 경량화 기법인 양자화도 적용하였다. 적용 결과 학습시 요구되는 GPU의 사용량이 32GB에서 5.7GB로 82.19% 감소함을 확인하였다. 동일한 조건에서 동일한 데이터로 모델의 성능을 평가한 결과 동일 학습 횟수에선 LoRa와 양자화가 적용된 모델의 오류가 기본 모델보다 53.34% 증가함을 확인하였다. 모델 성능의 감소를 줄이기 위해서는 학습 횟수를 더 증가시킨 결과 오류 증가율이 29.29%로 동일 학습 횟수보다 더 줄어듬을 확인하였다.

• Journal of the Korean Vacuum Society
• /
• v.4 no.3
• /
• pp.327-333
• /
• 1995

전자소자의 축소화에 따라 박막배선에서의 electromigration은 점차 극소전자 디바이스의 주요 결함원인으로 부각되고 있다. 본 실험에서는 현재 박막 배선 재료로 가장 널리 사용되고 있는 AI-1%Si 금속박막배선의 electromigration에 대한 온도 및 배선길이의 의존성에 관하여 연구하였다. PSG($8000AA$)/SiO2(1000$\AA$)/AI-1%Si(7000$\AA$)/SiO2(5000$\AA$)/p-Si(100)의 보호막처리되지 않은 시편 등을 standard photolithography 공정을 이용하여 각각 제작하였다. 선폭 3$\mu$m, 길이 100, 400, 800, $\1600mu$m등의 AI-1%Si 배막배선구조를 사용하였다. 가속화실험을 위해 인가된 d.c.전류밀도는 4.5X106A/$ extrm{cm}^2$이었고 실온에서 $100^{\circ}C$까지의 분위기 온도에서 electromigration test를 진행하였다. 박막배선의 길에에 따른 MTF(Mean-Time-to-Failure)는 임계길이 이상에서 포화되는 경향을 보이며 이는 보호막층의 유무에 관계없이 나타난다. 선폭 $3\mu$m인 AI-1%Si 박막배선에서 임계길이는, 보호막처리된 시편은 $800\mu$m, 보호막처리되지 않은 시편은 $400\mu$m 배선길이에서 나타난다. 이러한 포화의 경향은 낮은 온도에서 더욱 명확해지는 특성을 보인다. 각 시편에서 electromigration에 대한 활성화에너지도 MTF의 특성과 유사하게 임계길이 이상에서 포화되는 특성을 보인다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.617-618
• /
• 2024

본 논문에서 핵심적으로 연구할 내용은 기존 논문에서 소개된 BERT-base 모델의 경량화 버전인 DistilBERT 모델을 임베디드 시스템(Raspberry PI 5) 환경에 탑재 및 구현하는 것이다. 또한, 본 논문에서는 임베디드 시스템(Raspberry PI 5) 환경에 탑재한 DistilBERT 모델과 BERT-base 모델 간의 성능 비교를 수행하였다. 성능 평가에 사용한 데이터셋은 SQuAD(Standford Question Answering Dataset)로 질의응답 태스크에 대한 데이터셋이며, 성능 검증 지표로는 EM(Exact Match) Score와 F1 Score 그리고 추론시간을 사용하였다. 실험 결과를 통해 DistilBERT와 같은 경량화 모델이 임베디드 시스템(Raspberry PI 5)과 같은 환경에서 온 디바이스 AI(On-Device AI)로 잘 작동함을 증명하였다.

• Journal of the Korea Computer Graphics Society
• /
• v.30 no.3
• /
• pp.159-169
• /
• 2024

This paper proposes a new real-time fire monitoring and evacuation navigation system by integrating Augmented Reality (AR) technology with Internet of Things (IoT) technology. The proposed system collects temperature data through IoT temperature measurement devices installed in buildings and automatically transmits it to a MySQL cloud database via an IoT platform, enabling real-time and accurate data monitoring. Subsequently, the real-time IoT data is visualized on a 3D building model generated through Building Information Modeling (BIM), and the model is represented in the real world using AR technology, allowing intuitive identification of the fire origin. Furthermore, by utilizing Vuforia engine's Device Tracking and Area Targets features, the system tracks the user's real-time location and employs an enhanced A^* algorithm to find the optimal evacuation route among multiple exits. The paper evaluates the proposed system's practicality and demonstrates its effectiveness in rapid and safe evacuation through user experiments based on various virtual fire scenarios.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.3
• /
• pp.151-158
• /
• 2024

Although on-device artificial intelligence (AI) has gained attention to diagnosing machine faults in real time, most previous studies did not consider the model retraining and redeployment processes that must be performed in real-world industrial environments. Our study addresses this challenge by proposing an on-device AI-based real-time machine fault diagnosis system that utilizes continual learning. Our proposed system includes a lightweight convolutional neural network (CNN) model, a continual learning algorithm, and a real-time monitoring service. First, we developed a lightweight 1D CNN model to reduce the cost of model deployment and enable real-time inference on the target edge device with limited computing resources. We then compared the performance of five continual learning algorithms with three public bearing fault datasets and selected the most effective algorithm for our system. Finally, we implemented a real-time monitoring service using an open-source data visualization framework. In the performance comparison results between continual learning algorithms, we found that the replay-based algorithms outperformed the regularization-based algorithms, and the experience replay (ER) algorithm had the best diagnostic accuracy. We further tuned the number and length of data samples used for a memory buffer of the ER algorithm to maximize its performance. We confirmed that the performance of the ER algorithm becomes higher when a longer data length is used. Consequently, the proposed system showed an accuracy of 98.7%, while only 16.5% of the previous data was stored in memory buffer. Our lightweight CNN model was also able to diagnose a fault type of one data sample within 3.76 ms on the Raspberry Pi 4B device.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.4
• /
• pp.165-172
• /
• 2020

In this paper we propose an uses on-device-based edge computing technology and big data analysis methods through the use of on-device-based edge computing technology and analysis of big data, which are distributed computing paradigms that introduce computations and storage devices where necessary to solve problems such as transmission delays that occur when data is transmitted to central centers and processed in current general smart factories. However, even if edge computing-based technology is applied in practice, the increase in devices on the network edge will result in large amounts of data being transferred to the data center, resulting in the network band reaching its limits, which, despite the improvement of network technology, does not guarantee acceptable transfer speeds and response times, which are critical requirements for many applications. It provides the basis for developing into an AI-based facility prediction conservation analysis tool that can apply deep learning suitable for big data in the future by supporting intelligent facility management that can support productivity growth through research that can be applied to the field of facility preservation and smart factory industry with integrated hardware technology that can accommodate these requirements and factory management and control technology.