• The Korean Journal of Applied Statistics
• /
• v.36 no.2
• /
• pp.101-114
• /
• 2023

The data augmentation technique is effectively used to solve the problem of overfitting the model by allowing the training dataset to be viewed from various perspectives. In addition to image augmentation techniques such as rotation, cropping, horizontal flip, and vertical flip, occlusion-based data augmentation methods such as Cutmix and Cutout have been proposed. For models based on speech data, it is possible to use an occlusion-based data-based augmentation technique after converting a 1D speech signal into a 2D spectrogram. In particular, SpecAugment is an occlusion-based augmentation technique for speech spectrograms. In this study, we intend to compare and study data augmentation techniques that can be used in the problem of false-voice detection. Using data from the ASVspoof2017 and ASVspoof2019 competitions held to detect fake audio, a dataset applied with Cutout, Cutmix, and SpecAugment, an occlusion-based data augmentation method, was trained through an LCNN model. All three augmentation techniques, Cutout, Cutmix, and SpecAugment, generally improved the performance of the model. In ASVspoof2017, Cutmix, in ASVspoof2019 LA, Mixup, and in ASVspoof2019 PA, SpecAugment showed the best performance. In addition, increasing the number of masks for SpecAugment helps to improve performance. In conclusion, it is understood that the appropriate augmentation technique differs depending on the situation and data.

• Journal of Internet Computing and Services
• /
• v.24 no.4
• /
• pp.25-36
• /
• 2023

Today, as AI (Artificial Intelligence) technology is introduced in various fields, including security, the development of technology is accelerating. However, with the development of AI technology, attack techniques that cleverly bypass malicious behavior detection are also developing. In the classification process of AI models, an Adversarial attack has emerged that induces misclassification and a decrease in reliability through fine adjustment of input values. The attacks that will appear in the future are not new attacks created by an attacker but rather a method of avoiding the detection system by slightly modifying existing attacks, such as Adversarial attacks. Developing a robust model that can respond to these malware variants is necessary. In this paper, we propose two methods of generating Adversarial attacks as efficient Adversarial attack generation techniques for improving Robustness in AI models. The proposed technique is the XAI-based attack technique using the XAI technique and the Reference based attack through the model's decision boundary search. After that, a classification model was constructed through a malicious code dataset to compare performance with the PGD attack, one of the existing Adversarial attacks. In terms of generation speed, XAI-based attack, and reference-based attack take 0.35 seconds and 0.47 seconds, respectively, compared to the existing PGD attack, which takes 20 minutes, showing a very high speed, especially in the case of reference-based attack, 97.7%, which is higher than the existing PGD attack's generation rate of 75.5%. Therefore, the proposed technique enables more efficient Adversarial attacks and is expected to contribute to research to build a robust AI model in the future.

• The Transactions of the Korea Information Processing Society
• /
• v.13 no.10
• /
• pp.481-491
• /
• 2024

Digital watermarking is a technique for embedding information into digital content. Digital watermarking has attracted attention as a technique to combat piracy and identify artificially generated content, but it is still not robust in various situations. In this paper, we propose a frequency conversion-based hierarchical watermarking technique capable of attack detection, error correction, and owner identification. By embedding attack detection and error correction signatures in hierarchical watermarking, the proposed scheme maintains invisibility and outperforms the existing methods in capacity and robustness. We also proposed a framework to evaluate the performance of the image quality and error correction according to the type of error correction signature and the number of signature embeddings. We compared the visual quality and error correction performance of the conventional model without error correction signature and the conventional model with hamming and BCH signatures. We compared the quality by the number of signature embeddings and found that the quality deteriorates as the number of embeddings increases but is robust to attacks. By analyzing the quality and error correction ability by error correction signature type, we found that hamming codes showed better error correction performance than BCH codes and 41.31% better signature restoration performance than conventional methods.

• Korean Journal of Food Science and Technology
• /
• v.32 no.3
• /
• pp.531-537
• /
• 2000

The simple microgel electrophoresis of single cells, a 'comet assay', on fruit seeds enabled the rapid identification of irradiated fruits by comparing the intact non-irradiated cells and the damaged cells of irradiated fruits. Grapes and plums were irradiated with 0.1, 0.5, 0.7, 1.0 kGy and strawberries, peaches, apples, and nectarines were irradiated with only 1.0 kGy. Seeds were isolated, crushed, and the suspended cells were embedded in an agarose layer. After lysis of the cells, they were subjected to microgel electrophoresis for 2 minutes, and then stained. The DNA radiation-induced fragmentation of all the fruits stretched and migrated out of the cells forming a tail toward the anode giving the appearance of a comet, while the undamaged cells appeared as intact nuclei without tails. Grape and plum seeds irradiated at 0.5 kGy and higher showed significant increases in tail length. With increasing the irradiation doses, longer extention of the DNA from the nucleus toward the anode was observed. Strawberry, peach, apple, and nectarine seeds irradiated with 1.0 kGy also showed the longer tails than non-irradiated ones. DNA comet assay as a rapid and inexpensive screening technique could be an officially validated method for the detection of irradiated fruits.

• Journal of Intelligence and Information Systems
• /
• v.16 no.1
• /
• pp.93-116
• /
• 2010

Ensemble approach is applied to the detection modeling of illegal cash accommodation (ICA) that is the well-known type of fraudulent usages of credit cards in far east nations and has not been addressed in the academic literatures. The performance of fraud detection model (FDM) suffers from the imbalanced data problem, which can be remedied to some extent using an ensemble of many classifiers. It is generally accepted that ensembles of classifiers produce better accuracy than a single classifier provided there is diversity in the ensemble. Furthermore, recent researches reveal that it may be better to ensemble some selected classifiers instead of all of the classifiers at hand. For the effective detection of ICA, we adopt ensemble size reduction technique that prunes the ensemble of all classifiers using accuracy and diversity measures. The diversity in ensemble manifests itself as disagreement or ambiguity among members. Data imbalance intrinsic to FDM affects our approach for ICA detection in two ways. First, we suggest the training procedure with over-sampling methods to obtain diverse training data sets. Second, we use some variants of accuracy and diversity measures that focus on fraud class. We also dynamically calculate the diversity measure-Forward Addition and Backward Elimination. In our experiments, Neural Networks, Decision Trees and Logit Regressions are the base models as the ensemble members and the performance of homogeneous ensembles are compared with that of heterogeneous ensembles. The experimental results show that the reduced size ensemble is as accurate on average over the data-sets tested as the non-pruned version, which provides benefits in terms of its application efficiency and reduced complexity of the ensemble.

• Korean Journal of Remote Sensing
• /
• v.35 no.5_2
• /
• pp.819-829
• /
• 2019

In the disaster of collapse, an immediate response is needed to prevent the damage from worsening, and damage area calculation, response and recovery plan should be established. This requires accurate detection of the damage affected area. This study performed the detection of the damaged area by using UAV which can respond quickly and in real-time to detect the collapse accident. The study area was selected as B-05 housing redevelopment area in Jung-gu, Ulsan, where the demolition of houses and apartments in progress as the redevelopment project began. This area resembles a collapsed state of the building, which clear changes before and after the demolition. UAV images were acquired on May 17 and July 9, 2019, respectively. The changing area was considered as the damaged area before and after the collapse of the building, and the changing area was detected using CVA (Change Vector Analysis) the Representative Change Detection Technique, and SLIC (Simple Linear Iterative Clustering) based superpixel algorithm. In order to accurately perform the detection of the damaged area, the uninterested area (vegetation) was firstly removed using ExG (Excess Green), Among the objects that were detected by change, objects that had been falsely detected by area were finally removed by calculating the minimum area. As a result, the accuracy of the detection of damaged areas was 95.39%. In the future, it is expected to be used for various data such as response and recovery measures for collapse accidents and damage calculation.

• Korean Journal of Remote Sensing
• /
• v.34 no.6_1
• /
• pp.1025-1032
• /
• 2018

Snow Cover is a form of precipitation that is defined by snow on the surface and is the single largest component of the cryosphere that plays an important role in maintaining the energy balance between the earth's surface and the atmosphere. It affects the regulation of the Earth's surface temperature. However, since snow cover is mainly distributed in area where human access is difficult, snow cover detection using satellites is actively performed, and snow cover detection in forest area is an important process as well as distinguishing between cloud and snow. In this study, we applied the Normalized Difference Snow Index (NDSI) and the Normalized Difference Vegetation Index (NDVI) to the geostationary satellites for the snow detection of forest area in existing polar orbit satellites. On the rest of the forest area, the snow cover detection using $R_{1.61{\mu}m}$ anomaly technique and NDSI was performed. As a result of the indirect validation using the snow cover data and the Visible Infrared Imaging Radiometer (VIIRS) snow cover data, the probability of detection (POD) was 99.95 % and the False Alarm Ratio (FAR) was 16.63 %. We also performed qualitative validation using the Himawari-8 Advanced Himawari Imager (AHI) RGB image. The result showed that the areas detected by the VIIRS Snow Cover miss pixel are mixed with the area detected by the research false pixel.

• Journal of Korea Society of Industrial Information Systems
• /
• v.27 no.2
• /
• pp.71-82
• /
• 2022

In order to replace the sampling tensile test of products produced in the tab welding process, which is one of the automotive battery manufacturing processes, vision inspectors are currently being developed and used. However, the vision inspection has the problem of inspection position error and the cost of improving it. In order to solve these problems, there are recent cases of applying deep learning technology. As one such case, this paper tries to examine the usefulness of applying Faster R-CNN, one of the deep learning technologies, to existing product inspection. The images acquired through the existing vision inspection machine are used as training data and trained using the Faster R-CNN ResNet101 V1 1024x1024 model. The results of the conventional vision test and Faster R-CNN test are compared and analyzed based on the test standards of 0% non-detection and 10% over-detection. The non-detection rate is 34.5% in the conventional vision test and 0% in the Faster R-CNN test. The over-detection rate is 100% in the conventional vision test and 6.9% in Faster R-CNN. From these results, it is confirmed that deep learning technology is very useful for detecting welding error of lead tabs in automobile batteries.

• Journal of Bio-Environment Control
• /
• v.31 no.3
• /
• pp.255-260
• /
• 2022

This study aimed to develop a service model that uses a deep learning algorithm for detecting diseases and pests in strawberries through image data. In addition, the pest detection performance of deep learning models was further improved by proposing segmented image data sets specialized in disease and pest symptoms. The CNN-based YOLO deep learning model was selected to enhance the existing R-CNN-based model's slow learning speed and inference speed. A general image data set and a proposed segmented image dataset was prepared to train the pest and disease detection model. When the deep learning model was trained with the general training data set, the pest detection rate was 81.35%, and the pest detection reliability was 73.35%. On the other hand, when the deep learning model was trained with the segmented image dataset, the pest detection rate increased to 91.93%, and detection reliability was increased to 83.41%. This study concludes with the possibility of improving the performance of the deep learning model by using a segmented image dataset instead of a general image dataset.

• Journal of Radiation Industry
• /
• v.17 no.1
• /
• pp.19-32
• /
• 2023

The amount of radioactive waste is expected to dramatically increase with decommissioning of nuclear power plants such as Kori-1, the first nuclear power plant in South Korea. Accurate nuclide analysis is necessary to manage the radioactive wastes safely, but research on verification of radionuclide analysis has yet to be well established. This study aimed to develop the technology that can verify the results of radionuclide analysis based on artificial intelligence. In this study, we propose an anomaly detection algorithm for inspecting the analysis error of radionuclide. We used the data from 'Updated Scaling Factors in Low-Level Radwaste' (NP-5077) published by EPRI (Electric Power Research Institute), and resampling was performed using SMOTE (Synthetic Minority Oversampling Technique) algorithm to augment data. 149,676 augmented data with SMOTE algorithm was used to train the artificial neural networks (classification and anomaly detection networks). 324 NP-5077 report data verified the performance of networks. The anomaly detection algorithm of radionuclide analysis was divided into two modules that detect a case where radioactive waste was incorrectly classified or discriminate an abnormal data such as loss of data or incorrectly written data. The classification network was constructed using the fully connected layer, and the anomaly detection network was composed of the encoder and decoder. The latter was operated by loading the latent vector from the end layer of the classification network. This study conducted exploratory data analysis (i.e., statistics, histogram, correlation, covariance, PCA, k-mean clustering, DBSCAN). As a result of analyzing the data, it is complicated to distinguish the type of radioactive waste because data distribution overlapped each other. In spite of these complexities, our algorithm based on deep learning can distinguish abnormal data from normal data. Radionuclide analysis was verified using our anomaly detection algorithm, and meaningful results were obtained.