• 한국기계가공학회지
• /
• 제19권7호
• /
• pp.7-15
• /
• 2020

Recently, the fourth industrial revolution has been presented as a new paradigm and additive manufacturing (AM) has become one of the most important topics. For this reason, process monitoring for each cross-sectional layer of additive metal manufacturing is important. Particularly, deep learning can train a machine to analyze, optimize, and repair defects. In this paper, image classification is proposed by learning images of defects in the metal cross sections using the convolution neural network (CNN) image labeling algorithm. Defects were classified into three categories: crack, porosity, and hole. To overcome a lack-of-data problem, the amount of learning data was augmented using a data augmentation algorithm. This augmentation algorithm can transform an image to 180 images, increasing the learning accuracy. The number of training and validation images was 25,920 (80 %) and 6,480 (20 %), respectively. An optimized case with a combination of fully connected layers, an optimizer, and a loss function, showed that the model accuracy was 99.7 % and had a success rate of 97.8 % for 180 test images. In conclusion, image labeling was successfully performed and it is expected to be applied to automated AM process inspection and repair systems in the future.

• 한국음향학회지
• /
• 제41권2호
• /
• pp.122-129
• /
• 2022

본 논문은 프레임 속도를 국부적으로 조절하는 데이터 증강을 이용하여 트랜스포머 기반 음성 인식기의 성능을 개선하는 방법을 제안한다. 먼저, 원래의 음성데이터에서 증강할 부분의 시작 시간과 길이를 랜덤으로 선택한다. 그 다음, 선택된 부분의 프레임 속도는 선형보간법을 이용하여 새로운 프레임 속도로 변경된다. 월스트리트 저널 및 LibriSpeech 음성데이터를 이용한 실험결과, 수렴 시간은 베이스라인보다 오래 걸리지만, 인식 정확도는 대부분의 경우에 향상됨을 보여주었다. 성능을 더욱 향상시키기 위하여 변경 부분의 길이 및 속도 등 다양한 매개변수를 최적화하였다. 제안 방법은 월스트리트 저널 및 LibriSpeech 음성 데이터에서 베이스라인과 비교하여 각각 11.8 % 및 14.9 %의 상대적 성능 향상을 보여주는 것으로 나타났다.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2023년도 학술발표회
• /
• pp.95-95
• /
• 2023

Common hydrological problems of developing countries include poor data management, insufficient measuring devices and ungauged watersheds, leading to small or unreliable data availability. This has greatly affected the adoption of artificial intelligence techniques for flood risk mitigation and damage control in several developing countries. While climate datasets have recorded resounding applications, but they exhibit more uncertainties than ground-based measurements. To encourage AI adoption in developing countries with small ground-based dataset, we propose data augmentation for regression tasks and compare performance evaluation of different AI models with and without data augmentation. More focus is placed on simple models that offer lesser computational cost and higher accuracy than deeper models that train longer and consume computer resources, which may be insufficient in developing countries. To implement this approach, we modelled and predicted streamflow data of the Asa River Watershed located in Ilorin, Kwara State Nigeria. Results revealed that adequate hyperparameter tuning and proper model selection improve streamflow prediction on small water dataset. This approach can be implemented in data-scarce regions to ensure timely flood intervention and early warning systems are adopted in developing countries.

• 한국인터넷방송통신학회논문지
• /
• 제24권1호
• /
• pp.49-54
• /
• 2024

데이터 불균형은 분류 문제에서 흔히 마주치는 문제로, 데이터셋 내의 클래스간 샘플 수의 현저한 차이에서 기인한다. 이러한 데이터 불균형은 일반적으로 분류 모델에서 과적합, 과소적합, 성능 지표의 오해 등의 문제를 야기한다. 이를 해결하기 위한 방법으로는 Resampling, Augmentation, 규제 기법, 손실 함수 조정 등이 있다. 본 논문에서는 손실 함수 조정에 대해 다루며 특히, 불균형 문제를 가진 Multi-Class 블랙박스 동영상 데이터에서 여러 구성의 손실 함수(Cross Entropy, Balanced Cross Entropy, 두 가지 Focal Loss 설정: 𝛼 = 1 및 𝛼 = Balanced, Asymmetric Loss)의 성능을 I3D, R3D_18 모델을 활용하여 비교하였다.

• Journal of information and communication convergence engineering
• /
• 제16권3호
• /
• pp.173-178
• /
• 2018

Accurate classification of cloud images is a challenging task. Almost all the existing methods rely on hand-crafted feature extraction. Their limitation is low discriminative power. In the recent years, deep learning with convolution neural networks (CNNs), which can auto extract features, has achieved promising results in many computer vision and image understanding fields. However, deep learning approaches usually need large datasets. This paper proposes a deep learning approach for classification of cloud image patches on small datasets. First, we design a suitable deep learning model for small datasets using a CNN, and then we apply data augmentation and dropout regularization techniques to increase the generalization of the model. The experiments for the proposed approach were performed on SWIMCAT small dataset with k-fold cross-validation. The experimental results demonstrated perfect classification accuracy for most classes on every fold, and confirmed both the high accuracy and the robustness of the proposed model.

• 인터넷정보학회논문지
• /
• 제19권4호
• /
• pp.35-44
• /
• 2018

Monitoring through Synthesis Aperture Radar (SAR) is responsible for marine safety from floating icebergs. However, there are limits to distinguishing between icebergs and ships in SAR images. Convolutional Neural Network (CNN) is used to distinguish the iceberg from the ship. The goal of this paper is to increase the accuracy of identifying icebergs from SAR images. The metrics for performance evaluation uses the log loss. The two-layer CNN model proposed in research of C.Bentes et al.[1] is used as a benchmark model and compared with the four-layer CNN model using data augmentation. Finally, the performance of the final CNN model using the VGG-16 pre-trained model is compared with the previous model. This paper shows how to improve the benchmark model and propose the final CNN model.

• 항공우주시스템공학회지
• /
• 제4권1호
• /
• pp.19-22
• /
• 2010

The accuracy and integrity of global navigation satellite systems (GNSS) can be improved by using GNSS augmentation systems. Large ionospheric spatial gradient, during ionosphere storm, is a major threat for using GNSS augmentation systems by increasing the spatial decorrelation between a reference system and users. Ionosphere decorrelation behavior can be analyzed by using dual frequency GPS data. GNSS receivers have their own biases, called inter-frequency bias (IFB) between dual(P1 and P2) frequencies and they must be accurately estimated before computing ionosphere delays. GPS network data in Korea is used to compute each receiver's IFB, and their estimation accuracy and variability are analyzed. IFB estimation methodology to apply for ionosphere gradient analysis is discussed.

• Journal of Electrical Engineering and Technology
• /
• 제13권1호
• /
• pp.485-492
• /
• 2018

In this paper, we propose facial expression recognition using CNN (Convolutional Neural Network), one of the deep learning technologies. The proposed structure has general classification performance for any environment or subject. For this purpose, we collect a variety of databases and organize the database into six expression classes such as 'expressionless', 'happy', 'sad', 'angry', 'surprised' and 'disgusted'. Pre-processing and data augmentation techniques are applied to improve training efficiency and classification performance. In the existing CNN structure, the optimal structure that best expresses the features of six facial expressions is found by adjusting the number of feature maps of the convolutional layer and the number of nodes of fully-connected layer. The experimental results show good classification performance compared to the state-of-the-arts in experiments of the cross validation and the cross database. Also, compared to other conventional models, it is confirmed that the proposed structure is superior in classification performance with less execution time.

• Journal of Multimedia Information System
• /
• 제8권1호
• /
• pp.31-34
• /
• 2021

The classical HMM is defined by a parameter triple �� = (��, A, B), where each parameter represents a collection of probability distributions: initial state, state transition and output distributions in order. This paper proposes a new stationary parameter e = (e1, e2, …, eN) where N is the number of states and et = P(|xt = i, y) for describing how an input pattern y ends in state xt = i at time t followed by nothing. It is often said that all is well that ends well. We argue here that all should end well. The paper sets the framework for the theory and presents an efficient inference and training algorithms based on dynamic programming and expectation-maximization. The proposed model is applicable to analyzing any sequential data with two or more finite segmental patterns are concatenated, each forming a context to its neighbors. Experiments on online Hangul handwriting characters have proven the effect of the proposed augmentation in terms of highly intuitive segmentation as well as recognition performance and 13.2% error rate reduction.

• 대한임베디드공학회논문지
• /
• 제17권5호
• /
• pp.289-296
• /
• 2022

In this paper, we developed a deep learning-based recyclable object detection model. The model is developed based on YOLOv5 that is a one-stage detector. The deep learning model detects and classifies the recyclable object into 7 categories: paper, carton, can, glass, pet, plastic, and vinyl. We propose two methods for recyclable object detection models to solve problems during training. Bounding Box CutMix solved the no-objects training images problem of Mosaic, a data augmentation used in YOLOv5. Standardized Distance-based IoU replaced DIoU using a normalization factor that is not affected by the center point distance of the bounding boxes. The recyclable object detection model showed a final mAP performance of 0.91978 with Bounding Box CutMix and 0.91149 with Standardized Distance-based IoU.