• Journal of the Korea Society of Computer and Information
• /
• v.28 no.10
• /
• pp.67-76
• /
• 2023

Although deep learning models are making innovative achievements in the field of computer vision, the problem of vulnerability to adversarial examples continues to be raised. Adversarial examples are attack methods that inject fine noise into images to induce misclassification, which can pose a serious threat to the application of deep learning models in the real world. In this paper, we propose a model that detects adversarial examples using differences in predictive values between edge-learned classification models and underlying classification models. The simple process of extracting the edges of the objects and reflecting them in learning can increase the robustness of the classification model, and economical and efficient detection is possible by detecting adversarial examples through differences in predictions between models. In our experiments, the general model showed accuracy of {49.9%, 29.84%, 18.46%, 4.95%, 3.36%} for adversarial examples (eps={0.02, 0.05, 0.1, 0.2, 0.3}), whereas the Canny edge model showed accuracy of {82.58%, 65.96%, 46.71%, 24.94%, 13.41%} and other edge models showed a similar level of accuracy also, indicating that the edge model was more robust against adversarial examples. In addition, adversarial example detection using differences in predictions between models revealed detection rates of {85.47%, 84.64%, 91.44%, 95.47%, and 87.61%} for each epsilon-specific adversarial example. It is expected that this study will contribute to improving the reliability of deep learning models in related research and application industries such as medical, autonomous driving, security, and national defense.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.12
• /
• pp.519-524
• /
• 2023

In hashing-based image retrieval, the hash code of a manipulated image is different from the original image, making it difficult to search for the same image. This paper proposes and evaluates a self-supervised deephashing model that generates perceptual hash codes from feature information such as texture, shape, and color of images. The comparison models are autoencoder-based variational inference models, but the encoder is designed with a fully connected layer, convolutional neural network, and transformer modules. The proposed model is a variational inference model that includes a SimAM module of extracting geometric patterns and positional relationships within images. The SimAM module can learn latent vectors highlighting objects or local regions through an energy function using the activation values of neurons and surrounding neurons. The proposed method is a representation learning model that can generate low-dimensional latent vectors from high-dimensional input images, and the latent vectors are binarized into distinguishable hash code. From the experimental results on public datasets such as CIFAR-10, ImageNet, and NUS-WIDE, the proposed model is superior to the comparative model and analyzed to have equivalent performance to the supervised learning-based deephashing model. The proposed model can be used in application systems that require low-dimensional representation of images, such as image search or copyright image determination.

• Korean Journal of Radiology
• /
• v.23 no.10
• /
• pp.949-958
• /
• 2022

Objective: To investigate the feasibility of using a deep learning-based analysis of auscultation data to predict significant stenosis of arteriovenous fistulas (AVF) in patients undergoing hemodialysis requiring percutaneous transluminal angioplasty (PTA). Materials and Methods: Forty patients (24 male and 16 female; median age, 62.5 years) with dysfunctional native AVF were prospectively recruited. Digital sounds from the AVF shunt were recorded using a wireless electronic stethoscope before (pre-PTA) and after PTA (post-PTA), and the audio files were subsequently converted to mel spectrograms, which were used to construct various deep convolutional neural network (DCNN) models (DenseNet201, EfficientNetB5, and ResNet50). The performance of these models for diagnosing ≥ 50% AVF stenosis was assessed and compared. The ground truth for the presence of ≥ 50% AVF stenosis was obtained using digital subtraction angiography. Gradient-weighted class activation mapping (Grad-CAM) was used to produce visual explanations for DCNN model decisions. Results: Eighty audio files were obtained from the 40 recruited patients and pooled for the study. Mel spectrograms of "pre-PTA" shunt sounds showed patterns corresponding to abnormal high-pitched bruits with systolic accentuation observed in patients with stenotic AVF. The ResNet50 and EfficientNetB5 models yielded an area under the receiver operating characteristic curve of 0.99 and 0.98, respectively, at optimized epochs for predicting ≥ 50% AVF stenosis. However, Grad-CAM heatmaps revealed that only ResNet50 highlighted areas relevant to AVF stenosis in the mel spectrogram. Conclusion: Mel spectrogram-based DCNN models, particularly ResNet50, successfully predicted the presence of significant AVF stenosis requiring PTA in this feasibility study and may potentially be used in AVF surveillance.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.2
• /
• pp.31-41
• /
• 2024

Deep learning models show excellent performance in tasks such as image classification and object detection in the field of computer vision, and are used in various ways in actual industrial sites. Recently, research on improving robustness has been actively conducted, along with pointing out that this deep learning model is vulnerable to hostile examples. A hostile example is an image in which small noise is added to induce misclassification, and can pose a significant threat when applying a deep learning model to a real environment. In this paper, we tried to confirm the robustness of the edge-learning classification model and the performance of the adversarial example detection model using it for adversarial examples of various algorithms. As a result of robustness experiments, the basic classification model showed about 17% accuracy for the FGSM algorithm, while the edge-learning models maintained accuracy in the 60-70% range, and the basic classification model showed accuracy in the 0-1% range for the PGD/DeepFool/CW algorithm, while the edge-learning models maintained accuracy in 80-90%. As a result of the adversarial example detection experiment, a high detection rate of 91-95% was confirmed for all algorithms of FGSM/PGD/DeepFool/CW. By presenting the possibility of defending against various hostile algorithms through this study, it is expected to improve the safety and reliability of deep learning models in various industries using computer vision.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.26 no.3
• /
• pp.281-301
• /
• 2024

The power cable tunnels which are part of the underground transmission line project, are constructed using the shield TBM method. The disc cutter among the shield TBM components plays an important role in breaking rock mass. Efficient tunnel construction is possible only when appropriate replacement occurs as the wear limit is reached or damage such as uneven wear occurs. A study was conducted to determine the wear conditions of disc cutter using a deep learning algorithm based on real-time measurement data of wear and rotation speed. Based on the results of full-scaled tunnelling tests, it was confirmed that measurement data was obtained differently depending on the wear conditions of disc cutter. Using real-time measurement data, an algorithm was developed to determine disc cutter wear characteristics based on a convolutional neural network model. Distributional patterns of data can be learned through CNN filters, and the performance of the model that can classify uniform wear and uneven wear through these pattern features.

• Journal of the Society of Disaster Information
• /
• v.20 no.1
• /
• pp.199-205
• /
• 2024

Purpose: This study aims to improve the recognition rate of Auto People Counting (APC) in accurately identifying and providing information on remaining evacuees in disaster-vulnerable facilities such as nursing homes to firefighting and other response agencies in the event of a disaster. Methods: In this study, a baseline model was established using CNN (Convolutional Neural Network) models to improve the algorithm for recognizing images of incoming and outgoing individuals through cameras installed in actual disaster-vulnerable facilities operating APC systems. Various algorithms were analyzed, and the top seven candidates were selected. The research was conducted by utilizing transfer learning models to select the optimal algorithm with the best performance. Results: Experiment results confirmed the precision and recall of Densenet201 and Resnet152v2 models, which exhibited the best performance in terms of time and accuracy. It was observed that both models demonstrated 100% accuracy for all labels, with Densenet201 model showing superior performance. Conclusion: The optimal algorithm applicable to APC among various artificial intelligence algorithms was selected. Further research on algorithm analysis and learning is required to accurately identify the incoming and outgoing individuals in disaster-vulnerable facilities in various disaster situations such as emergencies in the future.

• Korean Journal of Radiology
• /
• v.22 no.11
• /
• pp.1909-1917
• /
• 2021

Objective: Muscle quantity and quality can be measured with an automated system on CT. However, the effects of contrast phases on the muscle measurements have not been established, which we aimed to investigate in this study. Materials and Methods: Muscle quantity was measured according to the skeletal muscle area (SMA) measured by a convolutional neural network-based automated system at the L3 level in 89 subjects undergoing multiphasic abdominal CT comprising unenhanced phase, arterial phase, portal venous phase (PVP), or delayed phase imaging. Muscle quality was analyzed using the mean muscle density and the muscle quality map, which comprises normal and low-attenuation muscle areas (NAMA and LAMA, respectively) based on the muscle attenuation threshold. The SMA, mean muscle density, NAMA, and LAMA were compared between PVP and other phases using paired t tests. Bland-Altman analysis was used to evaluate the inter-phase variability between PVP and other phases. Based on the cutoffs for low muscle quantity and quality, the counts of individuals who scored lower than the cutoff values were compared between PVP and other phases. Results: All indices showed significant differences between PVP and other phases (p < 0.001 for all). The SMA, mean muscle density, and NAMA increased during the later phases, whereas LAMA decreased during the later phases. Bland-Altman analysis showed that the mean differences between PVP and other phases ranged -2.1 to 0.3 cm² for SMA, -12.0 to 2.6 cm² for NAMA, and -2.2 to 9.9 cm² for LAMA.The number of patients who were categorized as low muscle quantity did not significant differ between PVP and other phases (p ≥ 0.5), whereas the number of patients with low muscle quality significantly differed (p ≤ 0.002). Conclusion: SMA was less affected by the contrast phases. However, the muscle quality measurements changed with the contrast phases to greater extents and would require a standardization of the contrast phase for reliable measurement.

• Journal of Korea Water Resources Association
• /
• v.56 no.2
• /
• pp.75-89
• /
• 2023

As the Mekong River basin is a nationally shared river, it is difficult to collect precipitation data, and the quantitative and qualitative quality of the data sets differs from country to country, which may increase the uncertainty of hydrological analysis results. Recently, with the development of remote sensing technology, it has become easier to obtain grid-based precipitation products(GPPs), and various hydrological analysis studies have been conducted in unmeasured or large watersheds using GPPs. In this study, rainfall-runoff simulation in the Mekong River basin was conducted using the SWAT model, which is a quasi-distribution model with three satellite GPPs (TRMM, GSMaP, PERSIANN-CDR) and two GPPs (APHRODITE, GPCC). Four water level stations, Luang Prabang, Pakse, Stung Treng, and Kratie, which are major outlets of the main Mekong River, were selected, and the parameters of the SWAT model were calibrated using APHRODITE as an observation value for the period from 2001 to 2011 and runoff simulations were verified for the period form 2012 to 2013. In addition, using the ConvAE, a convolutional neural network model, spatio-temporal correction of original satellite precipitation products was performed, and rainfall-runoff performances were compared before and after correction of satellite precipitation products. The original satellite precipitation products and GPCC showed a quantitatively under- or over-estimated or spatially very different pattern compared to APHPRODITE, whereas, in the case of satellite precipitation prodcuts corrected using ConvAE, spatial correlation was dramatically improved. In the case of runoff simulation, the runoff simulation results using the satellite precipitation products corrected by ConvAE for all the outlets have significantly improved accuracy than the runoff results using original satellite precipitation products. Therefore, the bias correction technique using the ConvAE technique presented in this study can be applied in various hydrological analysis for large watersheds where rain guage network is not dense.

• Journal of the Korean Society of Radiology
• /
• v.83 no.2
• /
• pp.344-359
• /
• 2022

Purpose To develop a denoising convolutional neural network-based image processing technique and investigate its efficacy in diagnosing breast cancer using low-dose mammography imaging. Materials and Methods A total of 6 breast radiologists were included in this prospective study. All radiologists independently evaluated low-dose images for lesion detection and rated them for diagnostic quality using a qualitative scale. After application of the denoising network, the same radiologists evaluated lesion detectability and image quality. For clinical application, a consensus on lesion type and localization on preoperative mammographic examinations of breast cancer patients was reached after discussion. Thereafter, coded low-dose, reconstructed full-dose, and full-dose images were presented and assessed in a random order. Results Lesions on 40% reconstructed full-dose images were better perceived when compared with low-dose images of mastectomy specimens as a reference. In clinical application, as compared to 40% reconstructed images, higher values were given on full-dose images for resolution (p < 0.001); diagnostic quality for calcifications (p < 0.001); and for masses, asymmetry, or architectural distortion (p = 0.037). The 40% reconstructed images showed comparable values to 100% full-dose images for overall quality (p = 0.547), lesion visibility (p = 0.120), and contrast (p = 0.083), without significant differences. Conclusion Effective denoising and image reconstruction processing techniques can enable breast cancer diagnosis with substantial radiation dose reduction.

• Journal of Intelligence and Information Systems
• /
• v.25 no.1
• /
• pp.1-19
• /
• 2019

Artificial intelligence (AI) is one of the main driving forces leading the Fourth Industrial Revolution. The technologies associated with AI have already shown superior abilities that are equal to or better than people in many fields including image and speech recognition. Particularly, many efforts have been actively given to identify the current technology trends and analyze development directions of it, because AI technologies can be utilized in a wide range of fields including medical, financial, manufacturing, service, and education fields. Major platforms that can develop complex AI algorithms for learning, reasoning, and recognition have been open to the public as open source projects. As a result, technologies and services that utilize them have increased rapidly. It has been confirmed as one of the major reasons for the fast development of AI technologies. Additionally, the spread of the technology is greatly in debt to open source software, developed by major global companies, supporting natural language recognition, speech recognition, and image recognition. Therefore, this study aimed to identify the practical trend of AI technology development by analyzing OSS projects associated with AI, which have been developed by the online collaboration of many parties. This study searched and collected a list of major projects related to AI, which were generated from 2000 to July 2018 on Github. This study confirmed the development trends of major technologies in detail by applying text mining technique targeting topic information, which indicates the characteristics of the collected projects and technical fields. The results of the analysis showed that the number of software development projects by year was less than 100 projects per year until 2013. However, it increased to 229 projects in 2014 and 597 projects in 2015. Particularly, the number of open source projects related to AI increased rapidly in 2016 (2,559 OSS projects). It was confirmed that the number of projects initiated in 2017 was 14,213, which is almost four-folds of the number of total projects generated from 2009 to 2016 (3,555 projects). The number of projects initiated from Jan to Jul 2018 was 8,737. The development trend of AI-related technologies was evaluated by dividing the study period into three phases. The appearance frequency of topics indicate the technology trends of AI-related OSS projects. The results showed that the natural language processing technology has continued to be at the top in all years. It implied that OSS had been developed continuously. Until 2015, Python, C ++, and Java, programming languages, were listed as the top ten frequently appeared topics. However, after 2016, programming languages other than Python disappeared from the top ten topics. Instead of them, platforms supporting the development of AI algorithms, such as TensorFlow and Keras, are showing high appearance frequency. Additionally, reinforcement learning algorithms and convolutional neural networks, which have been used in various fields, were frequently appeared topics. The results of topic network analysis showed that the most important topics of degree centrality were similar to those of appearance frequency. The main difference was that visualization and medical imaging topics were found at the top of the list, although they were not in the top of the list from 2009 to 2012. The results indicated that OSS was developed in the medical field in order to utilize the AI technology. Moreover, although the computer vision was in the top 10 of the appearance frequency list from 2013 to 2015, they were not in the top 10 of the degree centrality. The topics at the top of the degree centrality list were similar to those at the top of the appearance frequency list. It was found that the ranks of the composite neural network and reinforcement learning were changed slightly. The trend of technology development was examined using the appearance frequency of topics and degree centrality. The results showed that machine learning revealed the highest frequency and the highest degree centrality in all years. Moreover, it is noteworthy that, although the deep learning topic showed a low frequency and a low degree centrality between 2009 and 2012, their ranks abruptly increased between 2013 and 2015. It was confirmed that in recent years both technologies had high appearance frequency and degree centrality. TensorFlow first appeared during the phase of 2013-2015, and the appearance frequency and degree centrality of it soared between 2016 and 2018 to be at the top of the lists after deep learning, python. Computer vision and reinforcement learning did not show an abrupt increase or decrease, and they had relatively low appearance frequency and degree centrality compared with the above-mentioned topics. Based on these analysis results, it is possible to identify the fields in which AI technologies are actively developed. The results of this study can be used as a baseline dataset for more empirical analysis on future technology trends that can be converged.