• 전력전자학회논문지
• /
• 제27권6호
• /
• pp.507-514
• /
• 2022

This study proposes an optimal design process for a high-frequency transformer that has a large leakage inductance for dual-active-bridge converters. Notably, conventional design processes have large errors in designing leakage transformers because mathematically modeling the leakage inductance of such transformers is difficult. In this work, the geometric parameters of a shell-type transformer are identified, and finite element analysis(FEA) simulation is performed to determine the magnetization inductance, leakage inductance, and copper loss of various shapes of shell-type transformers. Regression models for magnetization and leakage inductances and copper loss are established using the simulation results and the machine learning technique. In addition, to improve the regression models' performance, the regression models are tuned by adding featured parameters that consider the physical characteristics of the transformer. With the regression models, optimal high-frequency transformer designs and the Pareto front (in terms of volume and loss) are determined using NSGA-II. In the Pareto front, a desirable optimal design is selected and verified by FEA simulation and experimentation. The simulated and measured leakage inductances of the selected design match well, and this result shows the validity of the proposed design process.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2023년도 추계학술발표대회
• /
• pp.540-542
• /
• 2023

Due to the serious issues posed by facial manipulation technologies, many researchers are becoming increasingly interested in the identification of face forgeries. The majority of existing face forgery detection methods leverage powerful data adaptation ability of neural network to derive distinguishing traits. These deep learning-based detection methods frequently treat the detection of fake faces as a binary classification problem and employ softmax loss to track CNN network training. However, acquired traits observed by softmax loss are insufficient for discriminating. To get over these limitations, in this study, we introduce a novel discriminative feature learning based on Vision Transformer architecture. Additionally, a separation-center loss is created to simply compress intra-class variation of original faces while enhancing inter-class differences in the embedding space.

• 한국융합학회논문지
• /
• 제11권3호
• /
• pp.19-26
• /
• 2020

본 논문에서는 학습된 사전 기반 초해상도 결과를 개선하기 위해 분석한 손실 영역을 기반으로 학습 데이터를 적용하는 방법을 제안하였다. 기존의 학습된 사전 기반 방법은 입력 영상의 특징을 고려하지 않는 학습된 영상의 형태로 출력할 수 있으며, 이 과정에서 인공물이 발생할 수 있다. 제안하는 방법은 입력 영상과 학습된 영상의 일치하지 않는 특징으로 인한 인공물 발생을 줄이기 위해 1차 복원 결과를 분석함으로써 손실 정보를 추정하였다. 추정된 결과의 잡음 및 화소 불균형을 가우시안 기반의 커널로 개선하여 생성된 특징 맵에 따라 학습 데이터를 매핑하였다. 결과 비교를 위해 기존의 초해상도 방법과 제안 방법의 결과를 고화질 영상과 PSNR(Peak Signal to Noise Ratio), SSIM(Structural SIMilarity Index) 으로 비교한 결과 각각 4%와 3%의 향상된 결과를 확인하였다.

• 대한한의학회지
• /
• 제41권2호
• /
• pp.58-79
• /
• 2020

Objectives: The purpose of this study is to predict the weight loss by applying machine learning using real-world clinical data from overweight and obese adults on weight loss program in 4 Korean Medicine obesity clinics. Methods: From January, 2017 to May, 2019, we collected data from overweight and obese adults (BMI≥23 kg/m2) who registered for a 3-month Gamitaeeumjowi-tang prescription program. Predictive analysis was conducted at the time of three prescriptions, and the expected reduced rate and reduced weight at the next order of prescription were predicted as binary classification (classification benchmark: highest quartile, median, lowest quartile). For the median, further analysis was conducted after using the variable selection method. The data set for each analysis was 25,988 in the first, 6,304 in the second, and 833 in the third. 5-fold cross validation was used to prevent overfitting. Results: Prediction accuracy was increased from 1^st to 2^nd and 3^rd analysis. After selecting the variables based on the median, artificial neural network showed the highest accuracy in 1^st (54.69%), 2^nd (73.52%), and 3^rd (81.88%) prediction analysis based on reduced rate. The prediction performance was additionally confirmed through AUC, Random Forest showed the highest in 1^st (0.640), 2^nd (0.816), and 3^rd (0.939) prediction analysis based on reduced weight. Conclusions: The prediction of weight loss by applying machine learning showed that the accuracy was improved by using the initial weight loss information. There is a possibility that it can be used to screen patients who need intensive intervention when expected weight loss is low.

• 한국재료학회지
• /
• 제32권8호
• /
• pp.345-353
• /
• 2022

In this study, using deep learning, super-resolution images of transmission electron microscope (TEM) images were generated for nanomaterial analysis. 1169 paired images with 256 × 256 pixels (high resolution: HR) from TEM measurements and 32 × 32 pixels (low resolution: LR) produced using the python module openCV were trained with deep learning models. The TEM images were related to DyVO₄ nanomaterials synthesized by hydrothermal methods. Mean-absolute-error (MAE), peak-signal-to-noise-ratio (PSNR), and structural similarity (SSIM) were used as metrics to evaluate the performance of the models. First, a super-resolution image (SR) was obtained using the traditional interpolation method used in computer vision. In the SR image at low magnification, the shape of the nanomaterial improved. However, the SR images at medium and high magnification failed to show the characteristics of the lattice of the nanomaterials. Second, to obtain a SR image, the deep learning model includes a residual network which reduces the loss of spatial information in the convolutional process of obtaining a feature map. In the process of optimizing the deep learning model, it was confirmed that the performance of the model improved as the number of data increased. In addition, by optimizing the deep learning model using the loss function, including MAE and SSIM at the same time, improved results of the nanomaterial lattice in SR images were achieved at medium and high magnifications. The final proposed deep learning model used four residual blocks to obtain the characteristic map of the low-resolution image, and the super-resolution image was completed using Upsampling2D and the residual block three times.

• 정보처리학회논문지:소프트웨어 및 데이터공학
• /
• 제9권2호
• /
• pp.61-68
• /
• 2020

e-sports는 최근 꾸준한 성장을 이루면서 세계적인 인기 스포츠 종목이 되었다. 본 논문에서는 e-sports의 대표적인 게임인 리그오브레전드 경기 시작 단계에서의 승패 예측 모델을 제안한다. 리그오브레전드에서는 챔피언이라고 불리는 게임 상의 유닛을 플레이어가 선택하여 플레이하게 되는데, 각 플레이어의 선택을 통하여 구성된 팀의 챔피언 능력치 조합은 승패에 영향을 미친다. 제안 모델은 별다른 도메인 지식 없이 플레이어 단위 챔피언 능력치를 팀 단위 챔피언 능력치로 임베딩한 Bidirectional LSTM 임베딩 기반 딥러닝 모델이다. 기존 분류 모델들과 비교 결과 팀 단위 챔피언 능력치 조합을 고려한 제안 모델에서 58.07%의 가장 높은 예측 정확도를 보였다.

• 한국인공지능학회지
• /
• 제5권1호
• /
• pp.29-37
• /
• 2017

The range of problems that can be handled by the activation of big data and the development of hardware has been rapidly expanded and machine learning such as deep learning has become a very versatile technology. In this paper, mnist data set is used as experimental data, and the Cross Entropy function is used as a loss model for evaluating the efficiency of machine learning, and the value of the loss function in the steepest descent method is We applied the Gradient Descent Optimize algorithm to minimize and updated weight and bias via backpropagation. In this way we analyze optimal reliability value corresponding to the number of exercises and optimal reliability value without overfitting. And comparing the overfitting time according to the number of data changes based on the number of training times, when the training frequency was 1110 times, we obtained the result of 92%, which is the optimal reliability value without overfitting.

• Journal of Electrical Engineering and Technology
• /
• 제13권5호
• /
• pp.2099-2106
• /
• 2018

This study presents a novel approach of discriminative feature vectors based on manifold learning using nonlinear dimension reduction (DR) technique to improve loss function, and combine with the Adversarial examples to regularize the object function for image classification. The traditional convolutional neural networks (CNN) with many new regularization approach has been successfully used for image classification tasks, and it achieved good results, hence it costs a lot of Calculated spacing and timing. Significantly, distrinct from traditional CNN, we discriminate the feature vectors for objects without empirically-tuned parameter, these Discriminative features intend to remain the lower-dimensional relationship corresponding high-dimension manifold after projecting the image feature vectors from high-dimension to lower-dimension, and we optimize the constrains of the preserving local features based on manifold, which narrow the mapped feature information from the same class and push different class away. Using Adversarial examples, improved loss function with additional regularization term intends to boost the Robustness and generalization of neural network. experimental results indicate that the approach based on discriminative feature of manifold learning is not only valid, but also more efficient in image classification tasks. Furthermore, the proposed approach achieves competitive classification performances for three benchmark datasets : MNIST, CIFAR-10, SVHN.

• 전기전자학회논문지
• /
• 제25권1호
• /
• pp.139-147
• /
• 2021

본 논문에서는 저조도 영상을 개선하기 위한 영상 분해 기반 심층 학습 방법 및 분해 채널 특성에 따른 손실함수를 제안한다. 기존 기법들의 문제점인 색신호 왜곡 및 할로 현상을 제거하기 위해, 입력 영상의 휘도 채널을 반사 성분과 조도 성분으로 분해하고, 반사 성분, 조도 성분 및 색차 신호를 신호 특성에 적합한 심층학습 과정을 적용하는 분해 기반 다중 구조 심층 학습 방법을 제안한다. 더불어, 분해 채널들의 특성에 따른 혼합 놈 기반의 손실함수를 정의하여 복원 영상의 안정성을 증대하고 열화 현상을 제거하기 위한 기법에 대해 기술한다. 실험 결과를 통해 제안한 방법이 다양한 저조도 영상을 효과적으로 개선하였음을 확인할 수 있었다.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제27권1호
• /
• pp.56-74
• /
• 2023

In order to obtain practical and high-quality satellite images containing high-frequency components, a large aperture optical system is required, which has a limitation in that it greatly increases the payload weight. As an attempt to overcome the problem, many multi-aperture optical systems have been proposed, but in many cases, these optical systems do not include high-frequency components in all directions, and making such an high-quality image is an ill-posed problem. In this paper, we use deep learning to overcome the limitation. A deep learning model receives low-quality images as input, estimates the Point Spread Function, PSF, and combines them to output a single high-quality image. We model images obtained from three rectangular apertures arranged in a regular polygon shape. We also propose the Modulation Transfer Function Loss, MTF Loss, which can capture the high-frequency components of the images. We present qualitative and quantitative results obtained through experiments.