• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2005.11a
• /
• pp.297-300
• /
• 2005

This paper presents vibration control of an engine mount for a passenger vehicle utilizing ER fluid and piezoelectric actuator. The proposed engine mount can be isolated the vibration of wide frequency range with many types of amplitude. The main function of ER fluid is to attenuate vibration for low frequency with large amplitude, while the piezoelectric actuator is activated in hish frequency range with small amplitude. A mathematical model of the engine mount is derived using Hydraulic model and mechanical model. After formulating the governing equation of motion, then field-dependent dynamic stiffness of the engine mount is evaluated for various engine speed and excitation amplitude conditions. Then robust controller is designed to attenuate vibration of wide range frequency component. Computer simulation is undertaken in order to evaluate the vibration control performance such as transmissibility magnitude in frequency domains.

• Journal of Cancer Prevention
• /
• v.22 no.4
• /
• pp.203-210
• /
• 2017

After transcription, RNAs are always associated with RNA binding proteins (RBPs) to perform biological activities. RBPs can interact with target RNAs in sequence- and structure-dependent manner through their unique RNA binding domains. In development and progression of carcinogenesis, RBPs are aberrantly dysregulated in many human cancers with various mechanisms, such as genetic alteration, epigenetic change, noncoding RNA-mediated regulation, and post-translational modifications. Upon deregulation in cancers, RBPs influence every step in the development and progression of cancer, including sustained cell proliferation, evasion of apoptosis, avoiding immune surveillance, inducing angiogenesis, and activating metastasis. To develop therapeutic strategies targeting RBPs, RNA interference-based oligonucleotides or small molecule inhibitors have been screened based on reduced RBP-RNA interaction and changed level of target RNAs. Identification of binding RNAs with high-throughput techniques and integral analysis of multiple datasets will help us develop new therapeutic drugs or prognostic biomarkers for human cancers.

• Wind and Structures
• /
• v.17 no.4
• /
• pp.399-417
• /
• 2013

The structural vibration suppression with active constrained layer damping (ACLD) was widely studied recently. However, the literature seldom concerned with the vibration control on flow-induced vibration using active constrained layer. In this paper the wind induced vibration of cantilevered beam is analyzed and suppressed by using random theory together with a velocity feedback control strategy. The piezoelectric material and frequency dependent viscoelastic layer are used to achieve effective active damping in the vibration control. The transverse displacement and velocity in time and frequency domains, as well as the power spectral density and the mean-square value of the transverse displacement and velocity, are formulated under wind pressure at variable control gain. It is observed from the numerical results that the wind induced vibration can be significantly suppressed by using a small outside active voltage on the constrained layer.

• Journal of the Korea Society of Computer and Information
• /
• v.23 no.1
• /
• pp.17-24
• /
• 2018

With improvement of speech recognition and natural language processing, dialog systems are recently adapted to various service domains. It became possible to get desirable services by conversation through the dialog system, but it is still necessary to improve separate modules, such as domain detection, intention detection, named entity recognition, and out-of-domain detection, in order to achieve stable service offer. When it misclassifies an in-domain sentence of conversation as out-of-domain, it will result in poor customer satisfaction and finally lost business. As there have been relatively small number of studies related to the out-of-domain detection, in this paper, we introduce a new method using a hierarchical Dirichlet process and demonstrate the effectiveness of it by experimental results on Korean dataset.

• ETRI Journal
• /
• v.44 no.3
• /
• pp.438-449
• /
• 2022

This study proposes a dependency grammar-based self-attention multilayered bidirectional long short-term memory (DG-M-Bi-LSTM) model for subject-predicate-object (SPO) tuple recognition from natural language (NL) sentences. To add recent knowledge to the knowledge base autonomously, it is essential to extract knowledge from numerous NL data. Therefore, this study proposes a high-accuracy SPO tuple recognition model that requires a small amount of learning data to extract knowledge from NL sentences. The accuracy of SPO tuple recognition using DG-M-Bi-LSTM is compared with that using NL-based self-attention multilayered bidirectional LSTM, DG-based bidirectional encoder representations from transformers (BERT), and NL-based BERT to evaluate its effectiveness. The DG-M-Bi-LSTM model achieves the best results in terms of recognition accuracy for extracting SPO tuples from NL sentences even if it has fewer deep neural network (DNN) parameters than BERT. In particular, its accuracy is better than that of BERT when the learning data are limited. Additionally, its pretrained DNN parameters can be applied to other domains because it learns the structural relations in NL sentences.

• International journal of advanced smart convergence
• /
• v.12 no.4
• /
• pp.164-170
• /
• 2023

With the recent advances in AI (artificial intelligence) and HPC (high-performance computing) technologies, deep learning is proliferated in various domains of the 4th industrial revolution. As the workload volume of deep learning increasingly grows, analyzing the memory reference characteristics becomes important. In this article, we analyze the memory reference traces of deep learning workloads in comparison with traditional workloads specially focusing on read and write operations. Based on our analysis, we observe some unique characteristics of deep learning memory references that are quite different from traditional workloads. First, when comparing instruction and data references, instruction reference accounts for a little portion in deep learning workloads. Second, when comparing read and write, write reference accounts for a majority of memory references, which is also different from traditional workloads. Third, although write references are dominant, it exhibits low reference skewness compared to traditional workloads. Specifically, the skew factor of write references is small compared to traditional workloads. We expect that the analysis performed in this article will be helpful in efficiently designing memory management systems for deep learning workloads.

• Journal of Environmental Science International
• /
• v.25 no.7
• /
• pp.1017-1028
• /
• 2016

Wind information is one of the major inputs for the prediction of urban air flow using computational fluid dynamic (CFD) models. Therefore, the numerical characteristics of the wind data formed at their mother domains should be clarified to predict the urban air flow more precisely. In this study, the formation characteristics of the wind data in the Seoul region were used as the inlet wind information for a CFD based simulation and were analyzed using numerical weather prediction models for weather research and forecasting (WRF). Because air flow over the central part of the Korean peninsula is often controlled not only by synoptic scale westerly winds but also by the westerly sea breeze induced from the Yellow Sea, the westerly wind often dominates the entire Seoul region. Although simulations of wind speed and air temperature gave results that were slightly high and low, respectively, their temporal variation patterns agreed well with the observations. In the analysis of the vertical cross section, the variation of wind speed along the western boundary of Seoul is simpler in a large domain with the highest horizontal resolution as compared to a small domain with the same resolution. A strong convergence of the sea breeze due to precise topography leads to the simplification of the wind pattern. The same tendency was shown in the average vertical profiles of the wind speed. The difference in the simulated wind pattern of two different domains is greater during the night than in the daytime because of atmospheric stability and topographically induced mesoscale forcing.

• KIEAE Journal
• /
• v.16 no.1
• /
• pp.81-87
• /
• 2016

Purpose: Heat transfers phenomena are described by the second order partial differential equation and its boundary conditions. In a three-dimensional structure of a building, the heat transfer phenomena generally include more than one material, and thus, become complicate. The analytic solutions are useful to understand heat transfer phenomena, but they can hardly be applied in engineering or design problems. Engineers and designers have generally been forced to use numerical methods providing reliable results. Finite volume methods with the unstructured grid system is only the suitable means of the analysis for the complex and arbitrary domains. Method: To obtain an numerical solution, a discretization method, which approximates the differential equations, and the interpolation methods for temperature and heat flux between two or more materials are required. The discretization methods are applied to small domains in space and time, and these numerical solutions form the descretized equations provide approximated solutions in both space and time. The accuracy of numerical solutions is dependent on the quality of discretizations and size of cells used. The higher accuracy, the higher numerical resources are required. The balance between the accuracy and difficulty of the numerical methods is critical for the success of the numerical analysis. A simple and easy interpolation methods among multiple materials are developed. The linear equations are solved with the BiCGSTAB being a effective matrix solver. Result: This study provides an overview of discretization methods, boundary interface, and matrix solver for the 3-dimensional numerical heat transfer including two materials.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.15 no.2
• /
• pp.211-223
• /
• 2005

The current software development type is based on CBD (Component Based Development). However, it requires a new component model that can support rapid and accurate component information on the web due to the cost increase of building new components using CBD. The purposed of this paper is to promptly develop an application on the web which fulfills users' demand and to provide specific interface regarding the small scales of distributed component in business domains. The e-BCOS(e-Business Component System) is the agent system for the users to register distributed components and to search component information, which would increase reusability through the e-business component development of distributed components in business domains and help the users understand the information about the component. Moreover, XML is employed for specification which allows smooth specification sharing and delivery to satisfy users' variable demands.

• The Mathematical Education
• /
• v.59 no.3
• /
• pp.217-235
• /
• 2020

This study examined Korean fourth-grade students' performance by gender on the Trends in International Mathematics and Science Study(TIMSS) 2011 and 2015 mathematics assessment. We first identified items which had significantly higher mean scores by gender to decide which gender did better on a certain domain(domain-level analysis). Then, we examined the content of items(item-level analysis) to understand which items lead to gender differences in mathematics achievement. Our findings showed that about 80% of the items on both assessments did not show statistically significant differences between males and females. However, there were meaningful gender differences in the other 20% items. On both assessments, females had more items with significantly higher mean scores than males on the Shapes domain, and males had more those items on the Numbers and Measurement domains and all cognitive domains(Knowing, Applying, and Reasoning). In particular, females outperformed males on items related to identifying two- and three-dimensional shapes and drawing lines and angles and identifying them. Conversely, males had higher performance than females on items related to the pre-algebraic thinking, fractions and decimals, estimation of number differences, unit of length, and measuring time, height, and volume. The effect sizes for each item ranged from .12 to .33 and the mean effect size of all items across both assessments was .20, which indicated significant gender differences but small.