• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.2
• /
• pp.74.2-75
• /
• 2021

We demonstrate that a deep learning classifier that only uses to gravitational wave (GW) detectors auxiliary channel data can distinguish various types of non-Gaussian noise transients (glitches) with significant accuracy, i.e., ≳ 80%. The classifier is implemented using the multi-scale neural networks (MSNN) with PyTorch. The glitches appearing in the GW strain data have been one of the main obstacles that degrade the sensitivity of the gravitational detectors, consequently hindering the detection and parameterization of the GW signals. Numerous efforts have been devoted to tracking down their origins and to mitigating them. However, there remain many glitches of which origins are not unveiled. We apply the MSNN classifier to the auxiliary channel data corresponding to publicly available GravitySpy glitch samples of LIGO O1 run without using GW strain data. Investigation of the auxiliary channel data of the segments that coincide to the glitches in the GW strain channel is particularly useful for finding the noise sources, because they record physical and environmental conditions and the status of each part of the detector. By only using the auxiliary channel data, this classifier can provide us with the independent view on the data quality and potentially gives us hints to the origins of the glitches, when using the explainable AI technique such as Layer-wise Relevance Propagation or GradCAM.

• Communications for Statistical Applications and Methods
• /
• v.30 no.6
• /
• pp.589-603
• /
• 2023

Deep generative models target to infer the underlying true data distribution, and it leads to a huge success in generating fake-but-realistic data. Regarding such a perspective, the data attributes can be a crucial factor in the data generation process since non-existent counterfactual samples can be generated by altering certain factors. For example, we can generate new portrait images by flipping the gender attribute or altering the hair color attributes. This paper proposes counterfactual disentangled variational autoencoder generative adversarial networks (CDVAE-GAN), specialized for data attribute level counterfactual data generation. The structure of the proposed CDVAE-GAN consists of variational autoencoders and generative adversarial networks. Specifically, we adopt a Gaussian variational autoencoder to extract low-dimensional disentangled data features and auxiliary Bernoulli latent variables to model the data attributes separately. Also, we utilize a generative adversarial network to generate data with high fidelity. By enjoying the benefits of the variational autoencoder with the additional Bernoulli latent variables and the generative adversarial network, the proposed CDVAE-GAN can control the data attributes, and it enables producing counterfactual data. Our experimental result on the CelebA dataset qualitatively shows that the generated samples from CDVAE-GAN are realistic. Also, the quantitative results support that the proposed model can produce data that can deceive other machine learning classifiers with the altered data attributes.

• The Journal of the Acoustical Society of Korea
• /
• v.27 no.8
• /
• pp.411-417
• /
• 2008

An alternative formulation of the Helmholtz integral equation derived to express the pressure field explicitly in terms of the velocity vector of a radiating surface is used to solve acoustic radiation and fluid/structure interaction problems. This formulation, derived for arbitrary sources, is similar in form to the Rayleigh's formula for planar sources. Because the surface pressure field is expressed explicitly as a surface integral of the surface velocity, which can be implemented numerically using standard Gaussian quadratures, there is no need to use BEM to solve a set of simultaneous equations for the surface pressure at the discretized nodes. Furthermore the non-uniqueness problem inherent in methods based on Helmholtz integral equation is avoided. Validation of this formulation is demonstrated for some simple geometries.

• Wind and Structures
• /
• v.38 no.4
• /
• pp.277-293
• /
• 2024

The recorded thunderstorm winds at a point contain tri-directional components. The probabilistic characteristics of such recorded winds in terms of instantaneous mean wind speed and direction, and the probability distribution and the time-frequency dependent crossed and non-crossed power spectral density functions for the high-frequency fluctuating wind components are unclear. In the present study, we analyze the recorded tri-directional thunderstorm wind components by separating the recorded winds in terms of low-frequency time-varying mean wind speed and high-frequency fluctuating wind components in the alongwind direction and two orthogonal crosswind directions. We determine the time-varying mean wind speed and direction defined by azimuth and elevation angles, and analyze the spectra of high-frequency wind components in three orthogonal directions using continuous wavelet transforms. Additionally, we evaluate the coherence between each pair of fluctuating winds. Based on the analysis results, we develop empirical spectral models and lagged coherence models for the tri-directional fluctuating wind components, and we indicate that the fluctuating wind components can be treated as Gaussian. We show how they can be used to generate time histories of the tri-directional thunderstorm winds.

• International Journal of Computer Science & Network Security
• /
• v.24 no.7
• /
• pp.143-147
• /
• 2024

Image noise classification is a classical problem in the field of image processing, machine learning, deep learning and computer vision. In this paper, image noise classification is performed using deep learning. Keras deep learning library of TensorFlow is used for this purpose. 6900 images images are selected from the Kaggle database for the classification purpose. Dataset for labeled noisy images of multiple type was generated with the help of Matlab from a dataset of non-noisy images. Labeled dataset comprised of Salt & Pepper, Gaussian and Sinusoidal noise. Different training and tests sets were partitioned to train and test the model for image classification. In deep neural networks CNN (Convolutional Neural Network) is used due to its in-depth and hidden patterns and features learning in the images to be classified. This deep learning of features and patterns in images make CNN outperform the other classical methods in many classification problems.

• Korean Journal of Radiology
• /
• v.25 no.7
• /
• pp.623-633
• /
• 2024

This study systematically reviewed the role of diffusion-weighted imaging (DWI) in the assessment of molecular prognostic biomarkers in breast cancer, focusing on the correlation of apparent diffusion coefficient (ADC) with hormone receptor status and prognostic biomarkers. Our meta-analysis includes data from 52 studies examining ADC values in relation to estrogen receptor (ER), progesterone receptor (PgR), human epidermal growth factor receptor 2 (HER2), and Ki-67 status. The results indicated significant differences in ADC values among different receptor statuses, with ER-positive, PgR-positive, HER2-negative, and Ki-67-positive tumors having lower ADC values compared to their negative counterparts. This study also highlights the potential of advanced DWI techniques such as intravoxel incoherent motion and non-Gaussian DWI to provide additional insights beyond ADC. Despite these promising findings, the high heterogeneity among the studies underscores the need for standardized DWI protocols to improve their clinical utility in breast cancer management.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.47 no.6
• /
• pp.28-38
• /
• 2010

In order to remove the noise contained in recorded analog video, it is important to recognize the real characteristics and strength of the noise. This paper presents an efficient training-based noise reduction method for recorded analog video after analyzing the noise characteristics of analog video captured in a real broadcasting system. First we show that there is non-negligible noise correlation in recorded analog video and describe the limitations of the traditional noise estimation and reduction methods based on additive white Gaussian noise (AWGN) model. In addition, we show that auto-regressive (AR) model considering noise correlation can be successfully utilized to estimate and synthesize the noise contained in the recorded analog video, and the estimated AR parameters are utilized in the training-based noise reduction scheme to reduce the video noise. Experiment results show that the proposed method can be efficiently applied for noise reduction of recorded analog video with non-negligible noise correlation.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.4 no.4
• /
• pp.201-207
• /
• 1992

The averaged non-linear term in the turbulence equations, suggested by Yeo (1987), is analyzed theoretically and experimentally. It was formulated by applying the filtering concepts to the convolution integral average definition with the Gaussian response function. This filtering approach seems to be superior to the conventional averaging methods in which all four terms at the doubly average vol must be defined separately, and it also gives a very useful tool in understanding the turbulence structures. By theoretically analyzing the newly derived description for the averaged non-linear terms, it is found that the vortex stretching can be explicitly accounted for. Furthermore, comparisons of the correlation coefficients based on the experimental data show that the vortex stretching acts most significantly on the turbulence residual stress. Thus, it strongly supports the claim that the vortex stretching is essential in the transfer of turbulence. In addition. a general form of turbulent energy models in LES is derived, by which it is recognized that the Smagorinsky, the vorticity and the SGS energy models are not distinctive.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• v.9 no.1
• /
• pp.413-416
• /
• 2005

Developement of computer and digitalizing of medical implement are being caused digitalizing of medical image. Medical Images are being managed more by PACS. In this situation comes out many problem about ownership, the right of properity of the medical images and notice wheather the data are modified. Digital watermarking can solve this problem. The proposed schemes have a two weak point that Non-blind method need a original image and Blind method use gaussian watermarking. In this paper, we propose a blind watermarking using binary image in order to easily recognize result of watermark This proposed scheme using wavelet transform and Circular Input method that input in medium-band of frequency domain.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2006.05a
• /
• pp.163-166
• /
• 2006

The field of medical images has been digitalized as the development of computer and the digitalization of the medical instruments. As a result it causes a lot of problems such as an illegal copy related to medical images and property right of the medical images. Therefore, digital watermarking is used for discrimination whether the data are modified or not. It is also used to protect both the property right of medical images and the private life of many patients. The proposed theories, the Non-blind and the Blind method, have two problems. One is needed an original image and the other is using a gaussian watermarking. This paper proposes the new Blind Watermarking using binary images in order to easily recognize the results of watermark. This algorithm is described that an watermark of a binary image is wavelet-transformed, and then a transformed watermark is inserted in medium-band of frequency domains of original image by the Circular Input method. The propose method presented the good performance of over 0.97 in NC.