• JSTS:Journal of Semiconductor Technology and Science
• /
• v.4 no.4
• /
• pp.269-274
• /
• 2004

Current delivery system, in which the analog current produced by a unique DAC circuit is stored into a current-memory circuit and delivered in a time-divided sequence, shows variation of output current as low as 4% in a current source data-driver IC for AM-OLED driven by a current-programmed method without any fuse repairing after fabrication. This driver IC has 54 outputs and can sink constant current as low as 3 ${\mu}A$ with 6-bit analog levels. Such a low current level without variation can hardly be obtained by an ordinary MOS transistor because the current level is in the sub-threshold region and changes exponentially with threshold voltage variation. Thus we adopted a current mirror circuit composed of bipolar transistors to supply well-controlled current within a nano-ampere range.

• Proceedings of the Korean Society for Bioinformatics Conference
• /
• 2005.09a
• /
• pp.237-242
• /
• 2005

Predicting the cellular location of an unknown protein gives a valuable information for inferring the possible function of the protein. For more accurate prediction system, we need a good feature extraction method that transforms the raw sequence data into the numerical feature vector, minimizing information loss. In this paper, we propose new methods of extracting underlying features only from the sequence data by computing pairwise sequence alignment scores. In addition, we use composition based features to improve prediction accuracy. To construct an SVM ensemble from separately trained SVM classifiers, we propose specificity based weighted majority voting. The overall prediction accuracy evaluated by the 5-fold cross-validation reached 88.53% for the eukaryotic animal data set. By comparing the prediction accuracy of various feature extraction methods, we could get the biological insight on the location of targeting information. Our numerical experiments confirm that our new feature extraction methods are very useful for predicting subcellular localization of proteins.

• Journal of the Korean Society for Aviation and Aeronautics
• /
• v.26 no.2
• /
• pp.39-46
• /
• 2018

Departure Manager (DMAN) is a tool to optimize the departure sequence and to suggest appropriate take-off time and off-block time of each departure aircraft to the air traffic controllers. To that end, Variable Taxi Time (VTT), which is time duration of the aircraft from the stand to the runway, should be estimated. In this paper, a study for development of VTT prediction algorithm based on machine learning techniques is presented. The factors affecting aircraft taxi speeds were identified through the analysis of historical traffic data on the airport surface. The prediction model suggested in this study consists of several sub-models that reflect different types of surface maneuvers based on the analysis result. The prediction performance of the proposed method was evaluated using the actual operational data.

• Journal of the Korea Society of Computer and Information
• /
• v.21 no.3
• /
• pp.97-104
• /
• 2016

In this paper, we propose spatial information based simulator for user experience optimization and minimize real space complexity. We focus on developing simulator how to design virtual space model and to implement virtual character using real space data. Especially, we use expanded events-driven inference model for SVM based on machine learning. Our simulator is capable of feature selection by k-fold cross validation method for optimization of data learning. This strategy efficiently throughput of executing inference of user behavior feature by virtual space model. Thus, we aim to develop the user experience optimization system for people to facilitate mapping as the first step toward to daily life data inference. Methodologically, we focus on user behavior and space modeling for implement virtual space.

• Annual Conference of KIPS
• /
• 2019.05a
• /
• pp.482-484
• /
• 2019

Learning and analyzing 3D point clouds with deep networks is challenging due to the limited and irregularity of the data. In this paper, we present a data-driven point cloud augmentation technique. The key idea is to learn multilevel features per point and to reconstruct to a similar point set. Our network is applied to a projection loss function that encourages the predicted points to remain on the geometric shapes with a particular target. We conduct various experiments using ShapeNet part data to evaluate our method and demonstrate its possibility. Results show that our generated points have a similar shape and are located closer to the object.

• Korean Journal of Remote Sensing
• /
• v.24 no.5
• /
• pp.463-471
• /
• 2008

The quality of orthoimages mainly depends on the elevation information and exterior orientation (EO) parameters. Since LiDAR data directly provides the elevation information over the earth's surface including buildings and trees, the concept of true orthorectification has been rapidly developed and implemented. If a LiDAR-driven digital surface model (DSM) is used for orthorectification, the displacements caused by trees and buildings are effectively removed when compared with the conventional orthoimages processed with a digital elevation model (DEM). This study utilized LiDAR data to generate orthorectified digital aerial images. Experimental orthoimages were produced using digital terrain model (DTM) and DSM. For the preparation of orthorectification, EO components, one of the inputs for orthorectification, were adjusted with the ground control points (GCPs) collected from the LiDAR point data, and the ground points were extracted by a filtering method used in a previous research. The orthoimage generated by DSM corresponded more closely to non-ground LiDAR points than the orthoimage produced by DTM.

• Journal of Ocean Engineering and Technology
• /
• v.36 no.3
• /
• pp.194-210
• /
• 2022

Recently, an analysis method using machine learning for solving problems in coastal and ocean engineering has been highlighted. Machine learning models are effective modeling tools for predicting specific parameters by learning complex relationships based on a specified dataset. In coastal and ocean engineering, various studies have been conducted to predict dependent variables such as wave parameters, tides, storm surges, design parameters, and shoreline fluctuations. Herein, we introduce and describe the application trend of machine learning models in coastal and ocean engineering. Based on the results of various studies, machine learning models are an effective alternative to approaches involving data requirements, time-consuming fluid dynamics, and numerical models. In addition, machine learning can be successfully applied for solving various problems in coastal and ocean engineering. However, to achieve accurate predictions, model development should be conducted in addition to data preprocessing and cost calculation. Furthermore, applicability to various systems and quantifiable evaluations of uncertainty should be considered.

• Proceedings of the KSRS Conference
• /
• v.2
• /
• pp.791-793
• /
• 2006

In the satellite camera, the incoming light source is converted to electronic analog signals by the electronic component for example CCD (Charge Coupled Device) detectors. The analog signals are amplified, biased and converted into digital signals (pixel data stream) in the video processor (A/Ds). The outputs of the A/Ds are digitally multiplexed and driven out using differential line drivers (two pairs of wires) for cross strap requirement. The MSC (Multi-Spectral Camera) in the KOMPSAT-2 which is a LEO spacecraft will be used to generate observation imagery data in two main channels. The MSC is to obtain data for high-resolution images by converting incoming light from the earth into digital stream of pixel data. The video data outputs are then MUXd, converted to 8 bit bytes, serialized and transmitted to the NUC (Non-Uniformity Correction) module by the Hotlink data transmitter. In this paper, the video data streams, the video data format, and the image data processing routine for satellite camera are described in terms of satellite camera control hardware. The advanced satellite with very high resolution requires faster and more complex image data chain than this algorithm. So, the effective change of the used image data chain and the fast video data transmission method are discussed in this paper

• KIPS Transactions on Software and Data Engineering
• /
• v.5 no.2
• /
• pp.69-78
• /
• 2016

Ischemic stroke is a disease that the brain tissues cannot function by reducing blood flow due to thrombosis or embolisms. Due to the nature of the disease, it is most important to identify the status of cerebral vessel and the medical images are necessarily used for its diagnosis. Among many indicators, brain MRI is most widely utilized because experts can effectively obtain the semantic information such as cerebral anatomy aiding the diagnosis with it. However, in case of emergency diseases like ischemic stroke, even though a intelligent system is required for supporting the prompt diagnosis and treatment, the current systems have some difficulties to provide the information of medical images intuitively. In other words, as the current systems have managed the medical images based on the basic meta-data such as image name, ID and so on, they cannot consider semantic information inherent in medical images. Therefore, in this paper, to provide core information like cerebral anatomy contained in brain MRI, we suggest a template-driven medical images mapping method. The key idea of the method is defining the mapping characteristics between anatomic feature and representative images by using template images that can be representative of the whole brain MRI image set and revealing the semantic relations that only medical experts can check between images. With our method, it will be possible to manage the medical images based on semantic.

• Smart Structures and Systems
• /
• v.10 no.4_5
• /
• pp.353-373
• /
• 2012

The instrumented Canton Tower is a 610 m high-rise structure, which has been considered as a benchmark problem for structural health monitoring (SHM) research. In this paper, an improved automatic modal identification method is presented based on a natural excitation technique in conjunction with the eigensystem realization algorithm (NExT/ERA). In the proposed modal identification method, damping ratio, consistent mode indicator from observability matrices (CMI_O) and modal amplitude coherence (MAC) are used as criteria to distinguish the physically true modes from spurious modes. Enhanced frequency domain decomposition (EFDD), the data-driven stochastic subspace identification method (SSI-DATA) and the proposed method are respectively applied to extract the modal parameters of the Canton Tower under different environmental conditions. Results of modal parameter identification based on output-only measurements are presented and discussed. User-selected parameters used in those methods are suggested and discussed. Furthermore, the effect of environmental conditions on the dynamic characteristics of Canton tower is investigated.