• Proceedings of the Korean Information Science Society Conference
• /
• 2000.10b
• /
• pp.308-310
• /
• 2000

관찰학습(OLA: Observational Learning Algorithm)은 앙상블 네트워크의 각 구성 모델들이 아른 모델들을 관찰함으로써 얻어진 가상 데이터와 초기에 bootstrap된 실제 데이터를 학습에 함께 이용하는 방법이다. 본 논문에서는, 초기 학습 데이터 셋을 분할하고 분할된 각 데이터 셋에 대하여 앙상블의 구성 모델들을 전문화(specialize)시키는 방법을 적용하여 기존의 관찰학습 알고리즘을 개선시켰다. 제안된 알고리즘은 bagging 및 boosting과의 비교 실험에 의하여, 보다 적은 수의 구성 모델로 동일 내지 보다 나은 성능을 나타냄이 실험적으로 검증되었다.

• Journal of the Korean Society of Propulsion Engineers
• /
• v.11 no.1
• /
• pp.43-50
• /
• 2007

In this study, Hybrid Separate Learning Algorithm(SLA) consisting of Support Vector Machine(SVM) and Artificial Neural Network(ANN) has been used for developing the defect diagnostic algorithm of the aircraft turbo-shaft engine in the off-design range considering altitude variation. Although the number of teaming data and test data highly increases more than 6 times compared with those required for the design condition, the proposed defect diagnostics of gas turbine engine using SLA was verified to give the high defect classification accuracy in the off-design range considering altitude variation.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10b
• /
• pp.422-424
• /
• 1999

본 논문에서는 질감 특징과 신경회로망을 이용한 유방암조직영상의 분할 방법을 제안한다. 신경회로망의 입력 노드에 사용될 질감 특징을 얻기 위해 10개의 영상에 대해 각 영역(양성세포핵, 음성세포핵, 배경)에서 10개씩의 화소를 선택하고, 그 화소를 중심으로 하는 5$\times$5 영역 30개를 획득, 총 300개의 영역에 대해 R, G, B 각각의 밴드에서 18개의 질감특징을 추출한다. 54개의 입력노드, 28개의 은닉노드, 3개의 출력노드의 구조를 가진 신경회로망을 구성하고, 역전파 학습 알고리즘을 사용하여 신경회로망을 최대오차율이 10-3보다 작을 때까지 학습시킨다. 학습에 의해 획득되어진 분류기를 이용하여 유방암 조직 세포영상을 양성세포핵, 음성세포핵, 배경부분으로 자동 분할한다.

• 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.

• Journal of the Korean Geotechnical Society
• /
• v.37 no.11
• /
• pp.107-118
• /
• 2021

Scan-to-BIM can be precisely mod eled by measuring structures with Light Detection And Ranging (LiDAR) and build ing a 3D BIM (Building Information Modeling) model based on it, but has a limitation in that it consumes a lot of manpower, time, and cost. To overcome these limitations, studies are being conducted to perform semantic segmentation of 3D point cloud data applying deep learning algorithms, but studies on how segmentation result changes depending on learning data are insufficient. In this study, a parametric study was conducted to determine how the size and track type of railroad tunnels constituting learning data affect the semantic segmentation of railroad tunnels through deep learning. As a result of the parametric study, the similar size of the tunnels used for learning and testing, the higher segmentation accuracy, and the better results when learning through a double-track tunnel than a single-line tunnel. In addition, when the training data is composed of two or more tunnels, overall accuracy (OA) and mean intersection over union (MIoU) increased by 10% to 50%, it has been confirmed that various configurations of learning data can contribute to efficient learning.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.10d
• /
• pp.616-618
• /
• 2002

본 논문에서는 음성인식 성능을 높이기 위한 기본적 단계인 음성과 비음성 부분의 경계를 추출하는 음성 경계 추출 방법을 제안한다. 음성경계 추출을 위한 특징들로는 시간영역 분할 파라미터인 ZCR, MA를 사용하고 주파수 영역 분할 파라미터로 주파수 대역 파워 에너지 (Frequency band power energy), 포만트 계수 (Formant coefficient)를 사용하였고 각 파라미터들을 이용하여 음성 경계를 결정할 때 경험에 의해 임계치를 결정하는 단점을 보안하기 위해서 신경망을 이용한다. 신경망의 가중치와 임계치들은 지도 학습을 통해 최적화 되고, 학습을 통해 구성된 망을 음성과 비음성의 경계치 구분에 사용한다.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_3
• /
• pp.967-978
• /
• 2023

This study proposed and validated a combined segmentation network (CSN) designed to effectively train on multiple drone image datasets and enhance the accuracy of semantic segmentation. CSN shares the entire encoding domain to accommodate the diversity of three drone datasets, while the decoding domains are trained independently. During training, the segmentation accuracy of CSN was lower compared to U-Net and the pyramid scene parsing network (PSPNet) on single datasets because it considers loss values for all dataset simultaneously. However, when applied to domestic autonomous drone images, CSN demonstrated the ability to classify pixels into appropriate classes without requiring additional training, outperforming PSPNet. This research suggests that CSN can serve as a valuable tool for effectively training on diverse drone image datasets and improving object recognition accuracy in new regions.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2000.10a
• /
• pp.33-36
• /
• 2000

관찰학습(OLA: Observational Learning Algorithm)은 앙상블 네트워크의 각 구성 모델들이 다른 모델들을 관찰함으로써 얻어진 가상 데이터와 초기에 bo otstrap된 실제 데이터를 학습에 함께 이용하는 방법이다. 본 논문에서는, 초기 학습 데이터 셋을 분할하고 분할된 각 데이터 셋에 대하여 앙상블의 구성 모델들을 전문화(specialize)시키는 방법을 적용하여 기존의 관찰학습 알고리즘을 개선시켰다. 제안된 알고리즘은 bagging 및 boosting과의 비교실험에 의하여, 보다 적은 수의 구성 모델로 동일 내지 보다 나은 성능을 나타냄이 실험적으로 검증되었다.

• Journal of IKEEE
• /
• v.25 no.1
• /
• pp.139-147
• /
• 2021

This paper introduces an image decomposition-based deep learning method and loss function to improve low-light images. In order to remove color distortion and halo artifact, illuminance channel of an input image is decomposed into reflectance and luminance channels, and a decomposition-based multiple structural deep learning process is applied to each channel. In addition, a mixed norm-based loss function is described to increase the stability and remove blurring in reconstructed image. Experimental results show that the proposed method effectively improve various low-light images.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.39 no.4
• /
• pp.462-469
• /
• 2002

While the conventional K-means algorithms use a fixed weight to design a vector codebook for all learning iterations, the proposed method employs a variable weight for learning iterations. The weight value of two or more beyond a convergent region is applied to obtain new codevectors at the initial learning iteration. The number of learning iteration applying a variable weight must be decreased for higher weight value at the initial learning iteration to design a better codebook. To enhance the splitting method that is used to generate an initial codebook, we propose a new method, which reduces the error between a representative vector and the member of training vectors. The method is that the representative vector with maximum squared error is rejected, but the vector with minimum error is splitting, and then we can obtain the better initial codevectors.