• 대한토목학회논문집
• /
• 제40권2호
• /
• pp.233-238
• /
• 2020

구조물 건전도 모니터링 시스템을 기반하는 교량 딥러닝 손상 추정 기법들은 대부분 지도학습을 기반으로 하고 있다. 지도학습의 특성상 손상 위치 추정 딥러닝 모델의 학습을 위해 교량의 손상 위치를 나타내는 라벨(Label) 데이터와 이에 따른 교량의 거동 데이터가 필요하다. 하지만 실제 현장에서 손상 위치 라벨 데이터를 정확히 얻어내는 것은 매우 어려운 일이므로, 지도학습 기반 딥러닝은 현장 적용성이 떨어진다는 한계가 있다. 반면에, 비지도학습 기반 딥러닝은 이러한 라벨 데이터 없이도 학습이 가능하다는 장점이 있다. 이러한 점에 착안하여 본 연구에서는 비지도 학습의 대표적인 딥러닝 기법인 Variational Autoencoder를 활용한 교량 손상 위치 추정의 방법을 제안하고 검증하였으며, 그 결과, 교량 손상 위치 추정을 위한 VAE의 적용 가능성을 보였다.

• 품질경영학회지
• /
• 제50권3호
• /
• pp.459-471
• /
• 2022

Purpose: The purpose of this study was to increase prediction accuracy for an anomaly interval identified using an artificial intelligence-based time series anomaly detection technique by establishing a pre-processing process. Methods: Significant variables were extracted by applying feature selection techniques, and anomalies were derived using the TadGAN time series anomaly detection algorithm. After applying machine learning and deep learning methodologies using normal section data (excluding anomaly sections), the explanatory power of the anomaly sections was demonstrated through performance comparison. Results: The results of the machine learning methodology, the performance was the best when SHAP and TadGAN were applied, and the results in the deep learning, the performance was excellent when Chi-square Test and TadGAN were applied. Comparing each performance with the papers applied with a Conventional methodology using the same data, it can be seen that the performance of the MLR was significantly improved to 15%, Random Forest to 24%, XGBoost to 30%, Lasso Regression to 73%, LSTM to 17% and GRU to 19%. Conclusion: Based on the proposed process, when detecting unsupervised learning anomalies of data that are not actually labeled in various fields such as cyber security, financial sector, behavior pattern field, SNS. It is expected to prove the accuracy and explanation of the anomaly detection section and improve the performance of the model.

• 한국컴퓨터그래픽스학회논문지
• /
• 제27권5호
• /
• pp.45-54
• /
• 2021

최근 노이즈 제거를 위한 심층 학습 모델에 대한 연구가 활발하게 진행되고 있다. 특히 블라인드 노이즈 제거 (blind denoising) 기술이 발전하면서 깨끗한 영상을 얻기가 불가능한 영상의 영역에서 노이즈 영상만으로 심층 학습 기반 노이즈 제거 모델의 학습이 가능해졌다. 우리는 관찰된 노이즈 영상으로부터 깨끗한 영상을 얻기 위해 더는 깨끗한 영상과 노이즈 영상의 짝을 이루는 데이터를 필요하지 않는다. 하지만 노이즈 영상과 깨끗한 영상 간의 차이가 큰 데이터라면 노이즈 영상만으로 학습된 노이즈 제거 모델은 우리가 원하는 품질의 깨끗한 영상을 복원하기 어려울 것이다. 이 문제를 해결하기 위해서 짝지어지지 않는 깨끗한 영상과 노이즈 영상으로 학습한 모델 기반 노이즈 제거 기술은 최근 연구되고 있다. 가장 최신 기술인 ISCL은 깨끗한 영상과 노이즈 영상의 쌍을 기반으로 한 지도학습 기반 모델의 성능과 거의 근접한 성능을 보여 주었다. 우리는 제안된 방법이 ISCL을 포함한 다른 최신 짝을 이루지 않는 영상 기반 노이즈 제거 기술보다 성능이 우수함을 보여준다.

• 로봇학회논문지
• /
• 제14권2호
• /
• pp.114-121
• /
• 2019

This paper presents a new benchmark system for visual odometry (VO) and monocular depth estimation (MDE). As deep learning has become a key technology in computer vision, many researchers are trying to apply deep learning to VO and MDE. Just a couple of years ago, they were independently studied in a supervised way, but now they are coupled and trained together in an unsupervised way. However, before designing fancy models and losses, we have to customize datasets to use them for training and testing. After training, the model has to be compared with the existing models, which is also a huge burden. The benchmark provides input dataset ready-to-use for VO and MDE research in 'tfrecords' format and output dataset that includes model checkpoints and inference results of the existing models. It also provides various tools for data formatting, training, and evaluation. In the experiments, the exsiting models were evaluated to verify their performances presented in the corresponding papers and we found that the evaluation result is inferior to the presented performances.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권10호
• /
• pp.5159-5178
• /
• 2018

Network Intrusion detection is a rapidly growing field of information security due to its importance for modern IT infrastructure. Many supervised and unsupervised learning techniques have been devised by researchers from discipline of machine learning and data mining to achieve reliable detection of anomalies. In this paper, a deep convolutional neural network (DCNN) based intrusion detection system (IDS) is proposed, implemented and analyzed. Deep CNN core of proposed IDS is fine-tuned using Randomized search over configuration space. Proposed system is trained and tested on NSLKDD training and testing datasets using GPU. Performance comparisons of proposed DCNN model are provided with other classifiers using well-known metrics including Receiver operating characteristics (RoC) curve, Area under RoC curve (AuC), accuracy, precision-recall curve and mean average precision (mAP). The experimental results of proposed DCNN based IDS shows promising results for real world application in anomaly detection systems.

• 한국멀티미디어학회논문지
• /
• 제25권8호
• /
• pp.991-998
• /
• 2022

Recently, object recognition using image/video signals is rapidly spreading on autonomous driving and mobile phones. However, the actual input image/video signals are easily exposed to a poor illuminance environment. A recent researches for improving illumination enable to estimate and compensate the illumination parameters. In this study, we propose VE-DCE (video enhancement zero-reference deep curve estimation) to improve the illumination of low-light images. The proposed VE-DCE uses unsupervised learning-based zero-reference deep curve, which is one of the latest among learning based estimation techniques. Experimental results show that the proposed method can achieve the quality of low-light video as well as images compared to the previous method. In addition, it can reduce the computational complexity with respect to the existing method.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2021년도 추계학술발표대회
• /
• pp.942-943
• /
• 2021

The present work proposes a classification system for the HEp-2 cell images using an unsupervised deep feature learning method. Unlike most of the state-of-the-art methods in the literature that utilize deep learning in a strictly supervised way, we propose here the use of the deep convolutional autoencoder (DCAE) as the principal feature extractor for classifying the different types of the HEp-2 cell images. The network takes the original cell images as the inputs and learns to reconstruct them in order to capture the features related to the global shape of the cells. A final feature vector is constructed by using the latent representations extracted from the DCAE, giving a highly discriminative feature representation. The created features will be fed to a nonlinear classifier whose output will represent the final type of the cell image. We have tested the discriminability of the proposed features on one of the most popular HEp-2 cell classification datasets, the SNPHEp-2 dataset and the results show that the proposed features manage to capture the distinctive characteristics of the different cell types while performing at least as well as the actual deep learning based state-of-the-art methods.

• 대한의용생체공학회:의공학회지
• /
• 제42권5호
• /
• pp.232-240
• /
• 2021

Heterogeneity in cancer is the major obstacle for precision medicine and has become a critical issue in the field of a cancer diagnosis. Many attempts were made to disentangle the complexity by molecular classification. However, multi-dimensional information from dynamic responses of cancer poses fundamental limitations on biomolecular marker-based conventional approaches. Cell morphology, which reflects the physiological state of the cell, can be used to track the temporal behavior of cancer cells conveniently. Here, we first present a hybrid learning-based platform that extracts cell morphology in a time-dependent manner using a deep convolutional neural network to incorporate multivariate data. Feature selection from more than 200 morphological features is conducted, which filters out less significant variables to enhance interpretation. Our platform then performs unsupervised clustering to unveil dynamic behavior patterns hidden from a high-dimensional dataset. As a result, we visualize morphology state-space by two-dimensional embedding as well as representative morphology clusters and trajectories. This cell morphology profiling strategy by hybrid learning enables simplification of the heterogeneous population of cancer.

• 한국컴퓨터정보학회:학술대회논문집
• /
• 한국컴퓨터정보학회 2020년도 제61차 동계학술대회논문집 28권1호
• /
• pp.11-14
• /
• 2020

The importance of data in the development of deep learning is very high. Data with high morphological features are usually utilized in the domains where careful lens calibrations are needed by a human to capture those data. Synthesis of high morphological data for that domain can be a great asset to improve the classification accuracy of systems in the field. Unsupervised learning can be employed for this task. Generating photo-realistic objects of interest has been massively studied after Generative Adversarial Network (GAN) was introduced. In this paper, we propose Morpho-GAN, a method that unifies several GAN techniques to generate quality data of high morphology. Our method introduces a new suitable training objective in the discriminator of GAN to synthesize images that follow the distribution of the original dataset. The results demonstrate that the proposed method can generate plausible data as good as other modern baseline models while taking a less complex during training.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송∙미디어공학회 2020년도 추계학술대회
• /
• pp.252-255
• /
• 2020

Unsupervised deep learning methods have shown impressive results for the challenging monocular depth estimation task, a field of study that has gained attention in recent years. A common approach for this task is to train a deep convolutional neural network (DCNN) via an image synthesis sub-task, where additional views are utilized during training to minimize a photometric reconstruction error. Previous unsupervised depth estimation networks are trained within a fixed depth estimation range, irrespective of its possible range for a given image, leading to suboptimal estimates. To overcome this suboptimal limitation, we first propose an unsupervised adaptive depth estimation method guided by minimum and maximum (min-max) depth priors for a given input image. The incorporation of min-max depth priors can drastically reduce the depth estimation complexity and produce depth estimates with higher accuracy. Moreover, we propose a novel network architecture for adaptive depth estimation, called the AdaMM-DepthNet, which adopts the min-max depth estimation in its front side. Intensive experimental results demonstrate that the adaptive depth estimation can significantly boost up the accuracy with a fewer number of parameters over the conventional approaches with a fixed minimum and maximum depth range.