• 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.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.06a
• /
• pp.572-573
• /
• 2022

본 논문은 메타버스 환경에서 문제가 대두되고있는 AI 윤리(ethic)를 배경으로 인터랙션을 통해 사람들의 온라인과 오프라인의 결정요소에 직접적으로 영향을 미치는 대화형 AI가 어떻게 윤리적으로 진화될 수 있을지에 대한 공학적 솔루션을 UX 관점으로 찾아보는 기술 전략 연구라고 할 수 있다. 연구의 가설은 AI 의 머신러닝과정에 개별 사용자 그룹의 경험데이터가 반드시 포함되고 고려되어야 AI 는 오류값을 줄이고 윤리적으로 대응할 수 있다는 전제이다. 이를 위하여 본 논문은 기존의 머신러닝과 대화형 AI 의 UX 관점의 다이아로그 플로우 등을 연구 분석하고 사용자 데이터들을 실험하여 메타버스 서비스 환경에서의 기존에 논의되고 있는 컨택스트기반의 AI 머신러닝 과정에 사용자의 정성적 경험데이터를 추가한 윤리적 UX 접근 개념 모델을 제안 하였다. 아직은 개념모델 단계이고 시스템에서는 지금까지 다르지 않았던 비정량적인 감정과 융합적경험을 어떻게 문화적으로 코드화 하고 시스템적인 랭귀지와 연결시킬 수 있을지에 대한사용자 연구가 후속연구로 진행될 예정이다.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.481-484
• /
• 2024

본 논문에서는 레고(LEGO®) 조립 설명서를 생성하기 위해 딥러닝을 이용한 조립 및 설명서 생성 시스템을 제안한다. 이 시스템은 사용자가 제공한 단일 이미지를 기반으로 레고 조립 설명서를 자동 생성한다. 해당 시스템은 딥러닝 기반 이미지 분할 기술을 활용하여 물체를 배경으로부터 분리하고 이를 통해 조립 설명서를 생성하는 과정을 포함하며, 조립을 위한 알고리즘을 새로 설계하였다. 이 시스템은 기존 레고 제품의 한계를 극복하고, 사용자에게 주어진 부품으로 다양한 모델을 자유롭게 조립할 수 있게 한다. 또한, 복잡한 레고 조립 과정을 간소화하고, 조립의 장벽을 낮추는 데 도움을 준다.

• The Journal of the Korea Contents Association
• /
• v.21 no.12
• /
• pp.13-22
• /
• 2021

The clinical decision support system uses accumulated medical data to apply an AI model learned by machine learning to patient diagnosis and treatment prediction. However, the existing black box-based AI application does not provide a valid reason for the result predicted by the system, so there is a limitation in that it lacks explanation. To compensate for these problems, this paper proposes a system model that applies XAI that can be explained in the development stage of the clinical decision support system. The proposed model can supplement the limitations of the black box by additionally applying a specific XAI technology that can be explained to the existing AI model. To show the application of the proposed model, we present an example of XAI application using LIME and SHAP. Through testing, it is possible to explain how data affects the prediction results of the model from various perspectives. The proposed model has the advantage of increasing the user's trust by presenting a specific reason to the user. In addition, it is expected that the active use of XAI will overcome the limitations of the existing clinical decision support system and enable better diagnosis and decision support.

• Annual Conference of KIPS
• /
• 2024.10a
• /
• pp.914-915
• /
• 2024

이 연구는 딥러닝을 활용하여 반려동물 얼굴을 인식하는 알고리즘을 개발하고, MobileNet, ResNet, DenseNet 등의 딥러닝 모델을 사용해 그 성능을 평가한다. 다양한 반려동물 이미지를 통해 각 모델의 학습 성능을 분석하고, 실종 반려동물 찾기 시스템에서 활용할 수 있는 최적의 모델을 제안하는 데 중점을 둔다. 연구 결과, MobileNet이 높은 정확도와 실시간 성능을 보여 반려동물 얼굴 인식 시스템에 적합한 모델로 평가되었다.

• Journal of Internet of Things and Convergence
• /
• v.10 no.5
• /
• pp.179-184
• /
• 2024

AI has the potential to enhance the effectiveness of early childhood education by offering personalized learning experiences, while AI-based educational tools are expected to improve teachers' work efficiency and contribute to the creation of innovative learning environments. Consequently, equipping pre-service early childhood teachers with the ability to adopt and effectively utilize AI technology is considered a crucial task in preparing for future educational environments. The Technology Acceptance Model (TAM) has been widely used to evaluate the acceptance of information technology services, but its limitations in predicting adoption intentions have been noted. To address these limitations, this study proposes an integrated model combining TAM with the Value-Based Adoption Model (VAM). A survey was conducted among pre-service early childhood teachers, and the collected data were analyzed using SPSS 27 and SmartPLS 4.0. The analysis revealed that perceived ease of use had a significant effect on perceived usefulness, and both perceived usefulness and enjoyment positively influenced AI adoption intention through the mediating effect of perceived value. Conversely, perceived risk was found to have a negative impact on AI adoption intention. The findings of this study are expected to deepen understanding of the potential for AI application in early childhood education and its acceptance, considering the broader implications of AI across various fields.

• Journal of Bio-Environment Control
• /
• v.33 no.1
• /
• pp.1-11
• /
• 2024

In this study, five artificial intelligence (AI) models: Inception v3, SqueezeNet (local), VGG-16, Painters, and DeepLoc were used to classify tea leaf diseases. Eight image categories were used: healthy, algal leaf spot, anthracnose, bird's eye spot, brown blight, gray blight, red leaf spot, and white spot. Software used in this study was Orange 3 which functions as a Python library for visual programming, that operates through an interface that generates workflows to visually manipulate and analyze the data. The precision of each AI model was recorded to select the ideal AI model. All models were trained using the Adam solver, rectified linear unit activation function, 100 neurons in the hidden layers, 200 maximum number of iterations in the neural network, and 0.0001 regularizations. To extend the functionality of Orange 3, new add-ons can be installed and, this study image analytics add-on was newly added which is required for image analysis. For the training model, the import image, image embedding, neural network, test and score, and confusion matrix widgets were used, whereas the import images, image embedding, predictions, and image viewer widgets were used for the prediction. Precisions of the neural networks of the five AI models (Inception v3, SqueezeNet (local), VGG-16, Painters, and DeepLoc) were 0.807, 0.901, 0.780, 0.800, and 0.771, respectively. Finally, the SqueezeNet (local) model was selected as the optimal AI model for the detection of tea diseases using tea leaf images owing to its high precision and good performance throughout the confusion matrix.

• Annual Conference of KIPS
• /
• 2021.05a
• /
• pp.382-384
• /
• 2021

AI (Artificial Intelligence)기술이 상용화되면서 최근 기업들은 AI 모델의 기능을 서비스화하여 제공하고 있다. 하지만 최근 이러한 서비스를 이용하여 기업이 자본을 투자해 학습시킨 AI 모델을 탈취하는 공격이 등장하여 위협이 되고 있다. 본 논문은 최근 연구되고 있는 이러한 모델 탈취 공격들에 대해 공격자의 정보를 기준으로 분류하여 서술한다. 또한 본 논문에서는 모델 탈취 공격에 대응하기 위해 다양한 관점에서 시도되는 방어 기법들에 대해 서술한다.

• Journal of Internet Computing and Services
• /
• v.24 no.4
• /
• pp.25-36
• /
• 2023

Today, as AI (Artificial Intelligence) technology is introduced in various fields, including security, the development of technology is accelerating. However, with the development of AI technology, attack techniques that cleverly bypass malicious behavior detection are also developing. In the classification process of AI models, an Adversarial attack has emerged that induces misclassification and a decrease in reliability through fine adjustment of input values. The attacks that will appear in the future are not new attacks created by an attacker but rather a method of avoiding the detection system by slightly modifying existing attacks, such as Adversarial attacks. Developing a robust model that can respond to these malware variants is necessary. In this paper, we propose two methods of generating Adversarial attacks as efficient Adversarial attack generation techniques for improving Robustness in AI models. The proposed technique is the XAI-based attack technique using the XAI technique and the Reference based attack through the model's decision boundary search. After that, a classification model was constructed through a malicious code dataset to compare performance with the PGD attack, one of the existing Adversarial attacks. In terms of generation speed, XAI-based attack, and reference-based attack take 0.35 seconds and 0.47 seconds, respectively, compared to the existing PGD attack, which takes 20 minutes, showing a very high speed, especially in the case of reference-based attack, 97.7%, which is higher than the existing PGD attack's generation rate of 75.5%. Therefore, the proposed technique enables more efficient Adversarial attacks and is expected to contribute to research to build a robust AI model in the future.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_4
• /
• pp.1111-1123
• /
• 2023

In recent times, semantic image segmentation methods using deep learning models have been widely used for monitoring changes in surface attributes using remote sensing imagery. To enhance the performance of various UNet-based deep learning models, including the prominent UNet model, it is imperative to have a sufficiently large training dataset. However, enlarging the training dataset not only escalates the hardware requirements for processing but also significantly increases the time required for training. To address these issues, transfer learning is used as an effective approach, enabling performance improvement of models even in the absence of massive training datasets. In this paper we present three transfer learning models, UNet-ResNet50, UNet-VGG19, and CBAM-DRUNet-VGG19, which are combined with the representative pretrained models of VGG19 model and ResNet50 model. We applied these models to building extraction tasks and analyzed the accuracy improvements resulting from the application of transfer learning. Considering the substantial impact of learning rate on the performance of deep learning models, we also analyzed performance variations of each model based on different learning rate settings. We employed three datasets, namely Kompsat-3A dataset, WHU dataset, and INRIA dataset for evaluating the performance of building extraction results. The average accuracy improvements for the three dataset types, in comparison to the UNet model, were 5.1% for the UNet-ResNet50 model, while both UNet-VGG19 and CBAM-DRUNet-VGG19 models achieved a 7.2% improvement.