• Smart Media Journal
• /
• v.11 no.2
• /
• pp.31-38
• /
• 2022

Promptly predicting changes in the salinity in rivers is an important task to predict the damage to agriculture and ecosystems caused by salinity infiltration and to establish disaster prevention measures. Because machine learning(ML) methods show much less computation cost than physics-based hydraulic models, they can predict the river salinity in a relatively short time. Due to shorter training time, ML methods have been studied as a complementary technique to physics-based hydraulic model. Many studies on salinity prediction based on machine learning have been studied actively around the world, but there are few studies in South Korea. With a massive number of datasets available publicly, we evaluated the performance of various kinds of machine learning techniques that predict the salinity of the Nakdong River Estuary Basin. As a result, LightGBM algorithm shows average 0.37 in RMSE as prediction performance and 2-20 times faster learning speed than other algorithms. This indicates that machine learning techniques can be applied to predict the salinity of rivers in Korea.

• Journal of IKEEE
• /
• v.26 no.1
• /
• pp.62-66
• /
• 2022

Recently, there have been many research efforts based on data-based deep learning technologies to deal with the interference problem between heterogeneous wireless communication devices in unlicensed frequency bands. However, existing approaches are commonly based on the use of complex neural network models, which require high computational power, limiting their efficiency in resource-constrained network interfaces and Internet of Things (IoT) devices. In this study, we address the problem of classifying heterogeneous wireless technologies including Wi-Fi and ZigBee in unlicensed spectrum bands. We focus on a data-driven approach that employs a supervised-learning method that uses received signal strength indicator (RSSI) data to train Deep Convolutional Neural Networks (CNNs). We propose a simple measurement methodology for collecting RSSI training data which preserves temporal and spectral properties of the target signal. Real experimental results using an open-source 2.4 GHz wireless development platform Ubertooth show that the proposed sampling method maintains the same accuracy with only a 10% level of sampling data for the same neural network architecture.

• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.2
• /
• pp.46.3-47
• /
• 2021

Recently a multilayer spectral inversion (MLSI) model has been proposed to infer the physical parameters of plasmas in the solar chromosphere. The inversion solves a three-layer radiative transfer model using the strong absorption line profiles, H alpha and Ca II 8542 Å, taken by the Fast Imaging Solar Spectrograph (FISS). The model successfully provides the physical plasma parameters, such as source functions, Doppler velocities, and Doppler widths in the layers of the photosphere to the chromosphere. However, it is quite expensive to apply the MLSI to a huge number of line profiles. For example, the calculating time is an hour to several hours depending on the size of the scan raster. We apply deep neural network (DNN) to the inversion code to reduce the cost of calculating the physical parameters. We train the models using pairs of absorption line profiles from FISS and their 13 physical parameters (source functions, Doppler velocities, Doppler widths in the chromosphere, and the pre-determined parameters for the photosphere) calculated from the spectral inversion code for 49 scan rasters (~2,000,000 dataset) including quiet and active regions. We use fully connected dense layers for training the model. In addition, we utilize a skip connection to avoid a problem of vanishing gradients. We evaluate the model by comparing the pairs of absorption line profiles and their inverted physical parameters from other quiet and active regions. Our result shows that the deep learning model successfully reproduces physical parameter maps of a scan raster observation per second within 15% of mean absolute percentage error and the mean squared error of 0.3 to 0.003 depending on the parameters. Taking this advantage of high performance of the deep learning model, we plan to provide the physical parameter maps from the FISS observations to understand the chromospheric plasma conditions in various solar features.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.33 no.3
• /
• pp.411-425
• /
• 2023

Today, AI (Artificial Intelligence) technology is widely used in various fields, performing classification and regression tasks according to the purpose of use, and research is also actively progressing. Especially in the field of security, unexpected threats need to be detected, and unsupervised learning-based anomaly detection techniques that can detect threats without adding known threat information to the model training process are promising methods. However, most of the preceding studies that provide interpretability for AI judgments are designed for supervised learning, so it is difficult to apply them to unsupervised learning models with fundamentally different learning methods. In addition, previously researched vision-centered AI mechanism interpretation studies are not suitable for application to the security field that is not expressed in images. Therefore, In this paper, we use a technique that provides interpretability for detected anomalies by searching for and comparing optimization references, which are the source of intrusion attacks. In this paper, based on reference, we propose additional logic to search for data closest to real data. Based on real data, it aims to provide a more intuitive interpretation of anomalies and to promote effective use of an anomaly detection model in the security field.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2020.06a
• /
• pp.125-125
• /
• 2020

European Space Agency's Sentinel-1 has an improved spatial and temporal resolution, as compared to previous satellite data such as Envisat Advanced SAR (ASAR) or Advanced Scatterometer (ASCAT). Thus, the assumption used for low-resolution retrieval algorithms used by ENVISAT ASAR or ASCAT is not applicable to Sentinel-1, because a higher degree of land surface heterogeneity should be considered for retrieval. The assumption of homogeneity over land surface is not valid any more. In this study, considering that soil roughness is one of the key parameters sensitive to soil moisture retrievals, various approaches are discussed. First, soil roughness is spatially inverted from Sentinel-1 backscattering over Yanco sites in Australia. Based upon this, Artificial Neural Networks data (feedforward multiplayer perception, MLP, Levenberg-Marquadt algorithm) are compared with Fractal approach (brownian fractal, Hurst exponent of 0.5). When using ANNs, training data are achieved from theoretical forward scattering models, Integral Equation Model (IEM). and Sentinel-1 measurements. The network is trained by 20 neurons and one hidden layer, and one input layer. On the other hand, fractal surface roughness is generated by fitting 1D power spectrum model with roughness spectra. Fractal roughness profile is produced by a stochastic process describing probability between two points, and Hurst exponent, as well as rms heights (a standard deviation of surface height). Main interest of this study is to estimate a spatial variability of roughness without the need of local measurements. This non-local approach is significant, because we operationally have to be independent from local stations, due to its few spatial coverage at the global level. More fundamentally, SAR roughness is much different from local measurements, Remote sensing data are influenced by incidence angle, large scale topography, or a mixing regime of sensors, although probe deployed in the field indicate point data. Finally, demerit and merit of these approaches will be discussed.

• The Bulletin of The Korean Astronomical Society
• /
• v.44 no.2
• /
• pp.55.3-55.3
• /
• 2019

We present a pipeline to estimate baryonic properties of a galaxy inside a dark matter (DM) halo in DM-only simulations using a machine trained on high-resolution hydrodynamic simulations. As an example, we use the IllustrisTNG hydrodynamic simulation of a (75 h^-1 Mpc)³ volume to train our machine to predict e.g., stellar mass and star formation rate in a galaxy-sized halo based purely on its DM content. An extremely randomized tree (ERT) algorithm is used together with multiple novel improvements we introduce here such as a refined error function in machine training and two-stage learning. Aided by these improvements, our model demonstrates a significantly increased accuracy in predicting baryonic properties compared to prior attempts --- in other words, the machine better mimics IllustrisTNG's galaxy-halo correlation. By applying our machine to the MultiDark-Planck DM-only simulation of a large (1 h^-1 Gpc)³ volume, we then validate the pipeline that rapidly generates a galaxy catalogue from a DM halo catalogue using the correlations the machine found in IllustrisTNG. We also compare our galaxy catalogue with the ones produced by popular semi-analytic models (SAMs). Our so-called machine-assisted semi-simulation model (MSSM) is shown to be largely compatible with SAMs, and may become a promising method to transplant the baryon physics of galaxy-scale hydrodynamic calculations onto a larger-volume DM-only run. We discuss the benefits that machine-based approaches like this entail, as well as suggestions to raise the scientific potential of such approaches.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.518-520
• /
• 2021

In this paper, we propose a hand gesture recognition method by modifying the textNAS used for text classification so that it can be applied to multivariate time series data. It can be applied to various fields such as behavior recognition, emotion recognition, and hand gesture recognition through multivariate time series data classification. In addition, it automatically finds a deep learning model suitable for classification through training, thereby reducing the burden on users and obtaining high-performance class classification accuracy. By applying the proposed method to the DHG-14/28 and Shrec'17 datasets, which are hand gesture recognition datasets, it was possible to obtain higher class classification accuracy than the existing models. The classification accuracy was 98.72% and 98.16% for DHG-14/28, and 97.82% and 98.39% for Shrec'17 14 class/28 class.

• Convergence Security Journal
• /
• v.22 no.5
• /
• pp.11-20
• /
• 2022

As the Internet and web technology develop around mobile devices, image data contains various types of sensitive information such as people, text, and space. In addition to these characteristics, as the use of SNS increases, the amount of damage caused by exposure and abuse of personal information online is increasing. However, research on de-identification technology based on multi-type object detection for personal information protection is insufficient. Therefore, this paper proposes an artificial intelligence model that detects and de-identifies multiple types of objects using existing single-type object detection models in parallel. Through cutmix, an image in which person and text objects exist together are created and composed of training data, and detection and de-identification of objects with different characteristics of person and text was performed. The proposed model achieves a precision of 0.724 and mAP@.5 of 0.745 when two objects are present at the same time. In addition, after de-identification, mAP@.5 was 0.224 for all objects, showing a decrease of 0.4 or more.

• Computers and Concrete
• /
• v.32 no.2
• /
• pp.217-232
• /
• 2023

Numerous studies have been performed on the behavior of pile foundations in cold regions. This study first attempted to employ artificial neural networks (ANN) to predict pile-bearing capacity focusing on pile data recorded primarily on cold regions. As the ANN technique has disadvantages such as finding global minima or slower convergence rates, this study in the second phase deals with the development of an ANN-based predictive model improved with an Elephant herding optimizer (EHO), Dragonfly Algorithm (DA), Genetic Algorithm (GA), and Evolution Strategy (ES) methods for predicting the piles' bearing capacity. The network inputs included the pile geometrical features, pile area (m²), pile length (m), internal friction angle along the pile body and pile tip (Ø°), and effective vertical stress. The MLP model pile's output was the ultimate bearing capacity. A sensitivity analysis was performed to determine the optimum parameters to select the best predictive model. A trial-and-error technique was also used to find the optimum network architecture and the number of hidden nodes. According to the results, there is a good consistency between the pile-bearing DA-MLP-predicted capacities and the measured bearing capacities. Based on the R2 and determination coefficient as 0.90364 and 0.8643 for testing and training datasets, respectively, it is suggested that the DA-MLP model can be effectively implemented with higher reliability, efficiency, and practicability to predict the bearing capacity of piles.

• Computers and Concrete
• /
• v.31 no.2
• /
• pp.111-121
• /
• 2023

The influence of material composition such as aggregate types, addition of supplementary cementitious materials as well as exposed temperature levels have significant impacts on concrete residual mechanical strength properties when exposed to elevated temperature. This study is based on data obtained from literature for fly ash blended concrete produced with natural and recycled concrete aggregates to efficiently develop prediction models for estimating its residual compressive strength after exposure to high temperatures. To achieve this, an extensive database that contains different mix proportions of fly ash blended concrete was gathered from published articles. The specific design variables considered were percentage replacement level of Recycled Concrete Aggregate (RCA) in the mix, fly ash content (FA), Water to Binder Ratio (W/B), and exposed Temperature level. Thereafter, a simplified mathematical equation for the prediction of concrete's residual compressive strength using Gene Expression Programming (GEP) was developed. The relative importance of each variable on the model outputs was also determined through global sensitivity analysis. The GEP model performance was validated using different statistical fitness formulas including R², MSE, RMSE, RAE, and MAE in which high R² values above 0.9 are obtained in both the training and validation phase. The low measured errors (e.g., mean square error and mean absolute error are in the range of 0.0160 - 0.0327 and 0.0912 - 0.1281 MPa, respectively) in the developed model also indicate high efficiency and accuracy of the model in predicting the residual compressive strength of fly ash blended concrete exposed to elevated temperatures.