• Proceedings of the Korea Water Resources Association Conference
• /
• 2015.05a
• /
• pp.225-225
• /
• 2015

Drought events usually evolve slowly in time and their impacts generally span a long period of time. This indicates that the sequence of drought is not completely random. The Hidden Markov Model (HMM) is a probabilistic model used to represent dependences between invisible hidden states which finally result in observations. Drought characteristics are dependent on the underlying generating mechanism, which can be well modelled by the HMM. This study employed a HMM with Gaussian emissions to fit the Standardized Precipitation Index (SPI) series and make multi-step prediction to check the drought characteristics in the future. To estimate the parameters of the HMM, we employed a Bayesian model computed via Markov Chain Monte Carlo (MCMC). Since the true number of hidden states is unknown, we fit the model with varying number of hidden states and used reversible jump to allow for transdimensional moves between models with different numbers of states. We applied the HMM to several stations SPI data in South Korea. The monthly SPI data from January 1973 to December 2012 was divided into two parts, the first 30-year SPI data (January 1973 to December 2002) was used for model calibration and the last 10-year SPI data (January 2003 to December 2012) for model validation. All the SPI data was preprocessed through the wavelet denoising and applied as the visible output in the HMM. Different lead time (T= 1, 3, 6, 12 months) forecasting performances were compared with conventional forecasting techniques (e.g., ANN and ARMA). Based on statistical evaluation performance, the HMM exhibited significant preferable results compared to conventional models with much larger forecasting skill score (about 0.3-0.6) and lower Root Mean Square Error (RMSE) values (about 0.5-0.9).

• Steel and Composite Structures
• /
• v.33 no.3
• /
• pp.319-332
• /
• 2019

This study is aimed to predict the behaviour of channel shear connectors in composite floor systems at different temperatures. For this purpose, a soft computing approach is adopted. Two novel intelligence methods, including an Extreme Learning Machine (ELM) and a Genetic Programming (GP), are developed. In order to generate the required data for the intelligence methods, several push-out tests were conducted on various channel connectors at different temperatures. The dimension of the channel connectors, temperature, and slip are considered as the inputs of the models, and the strength of the connector is predicted as the output. Next, the performance of the ELM and GP is evaluated by developing an Artificial Neural Network (ANN). Finally, the performance of the ELM, GP, and ANN is compared with each other. Results show that ELM is capable of achieving superior performance indices in comparison with GP and ANN in the case of load prediction. Also, it is found that ELM is not only a very fast algorithm but also a more reliable model.

• Korean Journal of Optics and Photonics
• /
• v.1 no.1
• /
• pp.46-51
• /
• 1990

Laser plasma was generated by a 1GW iodine photodissociation laser ($\lambda$=1.315$\mu\textrm{m}$, E=12.7J) whose output beam was focused on a molybdenum target surface. The experiment was conducted in a vacuum chamber under 1D-sTorr and several tens of laser shooting were necessary for sufficient exposure for the PBS resist of 111m thickness. Aluminium was coated on the top of the resist by 0.1$\mu\textrm{m}$ thickness which acts as an X-ray filter to cut off the visible and the ultraviolet lights. A bio-specimen was put directly on the aluminium coated resist and located at a distance of 3 cm from the X-ray source. The replicas of a steel mesh, spider's web. and a red blood cell were obtained by this technique and were observed by Nomarski microscope and SEM. The limitation of its resolution is determined by the X-ray source size and Fresnel diffraction effect, and its theoretical prediction is well matched with the experimental results. In this experiment, a resolution better than 0.1$\mu\textrm{m}$ could be obtained. ained.

• Journal of Navigation and Port Research
• /
• v.45 no.6
• /
• pp.284-297
• /
• 2021

A simple kinematic model for the prediction of ship manoeuvres based on trial data is proposed in this study. The model consists of first order differential equations in surge, sway, and yaw directions which simulate the time series of each velocity component. Actually instead of sea trial data, dynamic model simulations are conducted with randomly varied control inputs such as propeller revolution rates and rudder angles. Based on learning of control inputs and velocity outputs of dynamic model simulations in sufficient time, kinematic model coefficients are optimized so that the kinematic model can be approximately reproduce the velocity outputs of dynamic model simulations with arbitrary control inputs. The resultant kinematic model is verified with new dynamic simulation sets.

• Korean Journal of Radiology
• /
• v.19 no.6
• /
• pp.1021-1030
• /
• 2018

Objective: We evaluated the effect of various patient characteristics and time-density curve (TDC)-factors on the test bolus-affected vessel enhancement on coronary computed tomography angiography (CCTA). We also assessed the value of generalized linear regression models (GLMs) for predicting enhancement on CCTA. Materials and Methods: We performed univariate and multivariate regression analysis to evaluate the effect of patient characteristics and to compare contrast enhancement per gram of iodine on test bolus (${\Delta}HUTEST$) and CCTA (${\Delta}HUCCTA$). We developed GLMs to predict ${\Delta}HUCCTA$. GLMs including independent variables were validated with 6-fold cross-validation using the correlation coefficient and Bland-Altman analysis. Results: In multivariate analysis, only total body weight (TBW) and ${\Delta}HUTEST$ maintained their independent predictive value (p < 0.001). In validation analysis, the highest correlation coefficient between ${\Delta}HUCCTA$ and the prediction values was seen in the GLM (r = 0.75), followed by TDC (r = 0.69) and TBW (r = 0.62). The lowest Bland-Altman limit of agreement was observed with GLM-3 (mean difference, $-0.0{\pm}5.1$ Hounsfield units/grams of iodine [HU/gI]; 95% confidence interval [CI], -10.1, 10.1), followed by ${\Delta}HUCCTA$ ($-0.0{\pm}5.9HU/gI$; 95% CI, -11.9, 11.9) and TBW ($1.1{\pm}6.2HU/gI$; 95% CI, -11.2, 13.4). Conclusion: We demonstrated that the patient's TBW and ${\Delta}HUTEST$ significantly affected contrast enhancement on CCTA images and that the combined use of clinical information and test bolus results is useful for predicting aortic enhancement.

• The Journal of the Acoustical Society of Korea
• /
• v.37 no.6
• /
• pp.451-456
• /
• 2018

UAComm (Underwater Acoustic Communication) is an active research area, and many experiment has been performed to develop UAComm system. In this paper, we investigate the possibility of modifying and applying VirTEX (Virtual Time series EXperiment) to medium long range MIMO (Multiple-Input Multiple-Output) UAComm of about 20 km range for the analysis and performance prediction of UAComm system. Since VirTEX is a time-domain simulator, the generated time series can be used in HILS (Hardware In the Loop Simulation) to develop UAComm system. The developed package is verified through comparing with the sea-going FAF05 (Focused Acoustic Field 2005) experimental data. The developed simulator can be used to predict the performance of UAComm system, and even replace the expensive sea-going experiment.

• Journal of IKEEE
• /
• v.23 no.3
• /
• pp.852-857
• /
• 2019

In order to manage the demand resources of project participants and to provide appropriate strategies in the virtual power plant's power trading platform for consumers or operators who want to participate in the distributed resource collective trading market, it is very important to forecast the next day's demand of individual participants and the overall system's electricity demand. This paper developed a power demand forecasting model for the next day. For the model, we used LSTM algorithm of deep learning technique in consideration of time series characteristics of power demand forecasting data, and new scheme is applied by applying one-hot encoding method to input/output values such as power demand. In the performance evaluation for comparing the general DNN with our LSTM forecasting model, both model showed 4.50 and 1.89 of root mean square error, respectively, and our LSTM model showed high prediction accuracy.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.10
• /
• pp.3729-3749
• /
• 2021

At present, deep convolution network-based salient object detection (SOD) has achieved impressive performance. However, it is still a challenging problem to make full use of the multi-scale information of the extracted features and which appropriate feature fusion method is adopted to process feature mapping. In this paper, we propose a new adjacency auxiliary network (AANet) based on multi-scale feature fusion for SOD. Firstly, we design the parallel connection feature enhancement module (PFEM) for each layer of feature extraction, which improves the feature density by connecting different dilated convolution branches in parallel, and add channel attention flow to fully extract the context information of features. Then the adjacent layer features with close degree of abstraction but different characteristic properties are fused through the adjacent auxiliary module (AAM) to eliminate the ambiguity and noise of the features. Besides, in order to refine the features effectively to get more accurate object boundaries, we design adjacency decoder (AAM_D) based on adjacency auxiliary module (AAM), which concatenates the features of adjacent layers, extracts their spatial attention, and then combines them with the output of AAM. The outputs of AAM_D features with semantic information and spatial detail obtained from each feature are used as salient prediction maps for multi-level feature joint supervising. Experiment results on six benchmark SOD datasets demonstrate that the proposed method outperforms similar previous methods.

• The Journal of Asian Finance, Economics and Business
• /
• v.8 no.4
• /
• pp.83-91
• /
• 2021

Health is an outset of psychological, social, financial, and physical state. Several macroeconomic factors are entangled with health and mortality. Infant mortality and life expectancy are two keyguard on demographic research context on last few decades. On the other hand, foreign inflows play an unprecedent role for raising economic circulation and providing more opportunities to build a better society. The study aims to investigate the relationship between foreign direct investment (FDI), economic growth, and Bangladesh's health. This study employs time-series data from 1980 to 2018. Results show, with Auto-regressive Distribute Lag (ARDL) model, that there is significant cointegration among variables. Foreign investment and economic output relate significantly and positively to health. On the contrary, education is quasi-linked with a different sign-on different model. For model validation, pitfalls of time-series multicollinearity, heteroscedasiticy, and autocorrelation are not present. Also, CUSUM and CUSUMSQ tests are validating the model as stable and fit for future prediction. Medical assessment and education need more attention from the government as well as the private sector. FDI can play a catalyst role for improving the health sector, raising opportunity in educating and creating a better lifestyle. In order to optimize foreign investment, the government should implement necessary reforms and policies.

• Smart Media Journal
• /
• v.10 no.1
• /
• pp.32-38
• /
• 2021

Image classification through deep learning on the image from photographs has been a very active research field for the past several years. In this paper, we propose a method of automatically discriminating stone images from domestic source through deep learning, which is to use Python's hash library to scan 300×300 pixel photo images of granites such as Hwangdeungseok, Goheungseok, and Pocheonseok, performing data preprocessing to create learning images by examining duplicate images for each stone, removing duplicate images with the same hash value as a result of the inspection, and deep learning by stone. In addition, to utilize VGGNet, the size of the images for each stone is resized to 224×224 pixels, learned in VGG16 where the ratio of training and verification data for learning is 80% versus 20%. After training of deep learning, the loss function graph and the accuracy graph were generated, and the prediction results of the deep learning model were output for the three kinds of stone images.