• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2018.06a
• /
• pp.63-66
• /
• 2018

Recently, a convolutional neural network (CNN) models at single image super-resolution have been very successful. Residual learning improves training stability and network performance in CNN. In this paper, we compare four convolutional neural network models for super-resolution (SR) to learn nonlinear mapping from low-resolution (LR) input image to high-resolution (HR) target image. Four models include general CNN model, global residual learning CNN model, local residual learning CNN model, and the CNN model with global and local residual learning. Experiment results show that the results are greatly affected by how skip connections are connected at the basic CNN network, and network trained with only global residual learning generates highest performance among four models at objective and subjective evaluations.

• Proceedings of the IEEK Conference
• /
• 2000.11b
• /
• pp.128-131
• /
• 2000

In this paper, the animated simulation system (Anisim) is proposed in order to develope an efficient functional system verification tool. It displays the simulation results of the designed system using graphic animation with various models lot the target system. With simple interface definitions given by the user, Anisim generates interface codes automatically. Users can describe and model the target system with the generated interface codes. Since the simulation engine is implemented in C-language, modeling is very simple and simulation can be performed in real time.

• Journal of the military operations research society of Korea
• /
• v.13 no.1
• /
• pp.28-44
• /
• 1987

This thesis presents a model to assign targets and to determine gun ammunitions required for naval surface warfare. Delivery errors of weapon systems and vulnerability of moving targets are analyzed, then probability to kill moving battle sihps is computed. A weapon-target allocation model is proposed by using the Out of Kilter technique. A model to determine ammunitions required for killing moving targets is also designed. The models are evaluated by simulation and sensitivity analysis.

• Korean Management Science Review
• /
• v.32 no.4
• /
• pp.109-133
• /
• 2015

This paper considers the allocation and engagement scheduling of air defense missiles by using MIP (mixed integer programming). Specifically, it focuses on developing a realistic MIP model for a real battle situation where multiple enemy missiles are headed toward valuable defended assets and there exist multiple air defense missiles to counteract the threats. In addition to the conventional objective such as the minimization of surviving target value, the maximization of total intercept altitude is introduced as a new objective. The intercept altitude of incoming missiles is important in order to minimize damages from debris of the intercepted missiles and moreover it can be critical if the enemy warhead contains an atomic or chemical bomb. The concept of so called the time window is used to model the engagement situation and a continuous time is assumed for flying times of the both missiles. Lastly, the model is extended to simulate the situation where the guidance radar, which guides a defense missile to its target, has the maximum guidance capacity. The initial mathematical model developed contains several non-linear constraints and a non-linear objective function. Hence, the linearization of those terms is performed before it is solved by a commercially available software. Then to thoroughly examine the MIP model, the model is empirically evaluated with several test problems. Specifically, the models with different objective functions are compared and several battle scenarios are generated to evaluate performance of the models including the extended one. The results indicate that the new model consistently presents better and more realistic results than the compared models.

• The Korean Journal of Legal Medicine
• /
• v.41 no.2
• /
• pp.32-40
• /
• 2017

In addition to identifying genetic differences between target populations, it is also important to determine the impact of genetic differences with regard to the respective target populations. In recent years, there has been an increasing number of cases where this approach is needed, and thus various statistical methods must be considered. In this study, genetic data from populations of Southeast and Southwest Asia were collected, and several statistical approaches were evaluated on the Y-chromosome short tandem repeat data. In order to develop a more accurate and practical classification model, we applied gradient boosting and ensemble techniques. To infer between the Southeast and Southwest Asian populations, the overall performance of the classification models was better than that of the decision trees and regression models used in the past. In conclusion, this study suggests that additional statistical approaches, such as data mining techniques, could provide more useful interpretations for forensic analyses. These trials are expected to be the basis for further studies extending from target regions to the entire continent of Asia as well as the use of additional genes such as mitochondrial genes.

• Steel and Composite Structures
• /
• v.42 no.5
• /
• pp.633-647
• /
• 2022

This paper presents the results from reliability analyses of the current Eurocode 4 (EN 1994-1-1) and AISC 360-16 design models for predicting the resistance of headed stud shear connectors within profiled steel sheeting, when the ribs are oriented transverse to the supporting beam. For comparison purposes, the performance of the alternative "Luxembourg" and "Stuttgart" model were also considered. From an initial database of 611 push-out tests, 269 cases were included in the study, which ensured that the results were valid over a wide range of geometrical and material properties. It was found that the current EN 1994-1-1 design rules deliver a corrected partial safety factor γ_M^* of around 2.0, which is significantly higher than the target value 1.25. Moreover, 179 tests fell within the domain of the concrete-related failure design equation. Notwithstanding this, the EN 1994-1-1 equations provide satisfactory results for re-entrant profiled sheeting. The AISC 360-16 design equation for steel failure covers 263 of the tests in the database and delivers 𝛾_M^*≈2.0. Conversely, whilst the alternative "Stuttgart" model provides an improvement over the current codes, only a corrected partial safety factor of 𝛾_M^*=1.47 is achieved. Finally, the alternative "Luxembourg" design model was found to deliver the required target value, with a corrected partial safety factor 𝛾_M^* between 1.21 and 1.28. Given the fact that the Luxembourg design model is the only model that achieved the target values required by EN 1990, it is recommended as a potential candidate for inclusion within the second generation of Eurocodes.

• Journal of Information Technology Services
• /
• v.13 no.3
• /
• pp.221-233
• /
• 2014

Because people's interest of the stock market has been increased with the development of economy, a lot of studies have been going to predict fluctuation of stock prices. Latterly many studies have been made using scientific and technological method among the various forecasting method, and also data using for study are becoming diverse. So, in this paper we propose stock prices prediction models using sentiment analysis and machine learning based on news articles and SNS data to improve the accuracy of prediction of stock prices. Stock prices prediction models that we propose are generated through the four-step process that contain data collection, sentiment dictionary construction, sentiment analysis, and machine learning. The data have been collected to target newspapers related to economy in the case of news article and to target twitter in the case of SNS data. Sentiment dictionary was built using news articles among the collected data, and we utilize it to process sentiment analysis. In machine learning phase, we generate prediction models using various techniques of classification and the data that was made through sentiment analysis. After generating prediction models, we conducted 10-fold cross-validation to measure the performance of they. The experimental result showed that accuracy is over 80% in a number of ways and F1 score is closer to 0.8. The result can be seen as significantly enhanced result compared with conventional researches utilizing opinion mining or data mining techniques.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.18 no.3
• /
• pp.89-100
• /
• 2023

Recently, on-device artificial intelligence (AI) solutions using mobile devices and embedded edge devices have emerged in various fields, such as computer vision, to address network traffic burdens, low-energy operations, and security problems. Although vision transformer deep learning models have outperformed conventional convolutional neural network (CNN) models in computer vision, they require more computations and parameters than CNN models. Thus, they are not directly applicable to embedded edge devices with limited hardware resources. Many researchers have proposed various model compression methods or lightweight architectures for vision transformers; however, there are only a few studies evaluating the effects of model compression techniques of vision transformers on performance. Regarding this problem, this paper presents a performance evaluation of vision transformers on embedded platforms. We investigated the behaviors of three vision transformers: DeiT, LeViT, and MobileViT. Each model performance was evaluated by accuracy and inference time on edge devices using the ImageNet dataset. We assessed the effects of the quantization method applied to the models on latency enhancement and accuracy degradation by profiling the proportion of response time occupied by major operations. In addition, we evaluated the performance of each model on GPU and EdgeTPU-based edge devices. In our experimental results, LeViT showed the best performance in CPU-based edge devices, and DeiT-small showed the highest performance improvement in GPU-based edge devices. In addition, only MobileViT models showed performance improvement on EdgeTPU. Summarizing the analysis results through profiling, the degree of performance improvement of each vision transformer model was highly dependent on the proportion of parts that could be optimized in the target edge device. In summary, to apply vision transformers to on-device AI solutions, either proper operation composition and optimizations specific to target edge devices must be considered.

• Journal of the Korea Safety Management & Science
• /
• v.13 no.4
• /
• pp.193-199
• /
• 2011

In the past annual recycling obligation rate calculation of Extended Producer Responsibility (EPR) system, it was difficult to operate the system efficiently, because responsible producers passively participated in the scheme only bent on achieving annual obligation without long-term plan. Thus, a new scheme of long-term recycling obligation rate began to be established every five year from 2008 in order to give the basis for the notice of annual specific operation standard and recycling obligation, thereby helping responsible producers to make a preparation with a plan and giving expectation of active operation of the scheme. However, in the operation of long-term recycling target program, while the development of prediction models and the evaluation for existing items has been conducted in various ways, applications for a new target items and the evaluation are quite insufficient. Therefore, in this study, problems in implementing long-term recycling goal of new target items will be examined, and more objective and rational long-term recycling rate calculation and the operation standard will be proposed. Thus, the long-term recycling target will play a role as a pacemaker to steadily improve the recycling performance of target items, and responsible producers will be expected to increase the achievement with the realistic capacity.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.7
• /
• pp.34-42
• /
• 1999

Multi-target tracking system is defined as tracking several targets simultaneously. Kalman filter is widely used for target tracking problems. Kalman filter is known to be extremely useful as an optimal estimator but has a shortcoming of computational complexity. So a simplified estimator model which had less computational burden is proposed for a real-time implementation of multi-target tracking systems. In this paper, Kalman filter is applied to implement a real-time tracking system with a simplified target model. The proposed Kalman filter model is simpler compared with those of conventional ones, greatly reducing computation time, yet keeping the tracking abilities of the optimal Kalman filter. Through both simulations and experiments with real environments, it is demonstrated that the proposed simplified model works good in real situation with multiple to be tracked.