• Proceedings of the Korean Society of Computer Information Conference
• /
• 2023.07a
• /
• pp.81-83
• /
• 2023

GIT의 커밋 메시지를 소프트웨어 유지보수 활동 세 가지로 분류하는 연구를 분석하고 정확도를 높일 수 있는 모델들을 분석하였고 관련 모델 중 커밋메시지와 변경된 소스를 같이 활용하는 연구들은 변경된 소스를 분석하기 위해 도구들을 대부분 활용하는데 대부분 특정 언어만 분류할 수 있는 한계가 있다. 본 논문에서는 소스 변경 데이터를 추출할 때 언어의 제약을 없애기 위해 GPT를 이용해 변경된 소스의 요약을 추출하는 과정을 추가함으로써 언어 제약의 한계를 극복할 수 있는 개선된 모델에 관한 연구를 진행하였다. 향후 본 연구 모델의 구현 및 검증을 진행하고 이를 이용해 프로젝트 진행에 활용할 수 있는 솔루션 개발 연구까지 확정해 나갈 예정이다.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.143-148
• /
• 2023

본 논문에서는 금융 도메인 특화 사전학습 언어모델인 KF-DeBERTa(Korean Finance DeBERTa)를 제안한다. KF-DeBERTa는 대규모의 금융 말뭉치를 기반으로 학습하였으며, Transformer 아키텍처와 DeBERTa의 특징을 기반으로 구성되었다. 범용 및 금융 도메인에 대한 평가에서 KF-DeBERTa는 기존 언어모델들에 비해 상당히 높은 성능을 보였다. 특히, 금융 도메인에서의 성능은 매우 두드러졌으며, 범용 도메인에서도 다른 모델들을 상회하는 성능을 나타냈다. KF-DeBERTa는 모델 크기 대비 높은 성능 효율성을 보여주었고, 앞으로 금융 도메인에서의 활용도가 기대된다.

• Korean Journal of Radiology
• /
• v.25 no.2
• /
• pp.126-133
• /
• 2024

Large language models (LLMs) have revolutionized the global landscape of technology beyond natural language processing. Owing to their extensive pre-training on vast datasets, contemporary LLMs can handle tasks ranging from general functionalities to domain-specific areas, such as radiology, without additional fine-tuning. General-purpose chatbots based on LLMs can optimize the efficiency of radiologists in terms of their professional work and research endeavors. Importantly, these LLMs are on a trajectory of rapid evolution, wherein challenges such as "hallucination," high training cost, and efficiency issues are addressed, along with the inclusion of multimodal inputs. In this review, we aim to offer conceptual knowledge and actionable guidance to radiologists interested in utilizing LLMs through a succinct overview of the topic and a summary of radiology-specific aspects, from the beginning to potential future directions.

• Journal of the Semiconductor & Display Technology
• /
• v.23 no.1
• /
• pp.19-24
• /
• 2024

It is not an exaggeration to say that a single fire causes a lot of damage, so fires are one of the disaster situations that must be alerted as soon as possible. Various technologies have been utilized so far because preventing and detecting fires can never be completely accomplished with individual human efforts. Recently, deep learning technology has been developed, and fire detection systems using object detection neural networks are being actively studied. In this paper, we propose a new fire detection system that improves the previously studied fire detection system. We train the YOLOv8 model using refined datasets through improved labeling methods, derive results, and demonstrate the superiority of the proposed system by comparing it with the results of previous studies.

• Journal of Power Electronics
• /
• v.18 no.5
• /
• pp.1577-1584
• /
• 2018

The content of harmonic current increases with an increase in the number of power electronic devices in power grid. When a generator is directly connected to the power grid through a step-up transformer, the influence of the harmonic currents on the generator is inevitable. To study the influences of harmonics on generators, a 24-MW bulb tubular turbine generator is taken as an example in this paper. A 2-D transient electromagnetic field model is established. Through a comparative analysis of the data of experiments and simulations, the correctness of the model is verified. The values of the air gap magnetic density, torque and losses of the generator under various conditions are calculated using the finite element method. Taking the rated condition as a reference, the influence of the harmonic currents on the magnetic flux density is analyzed. It is confirmed that the time harmonic is a key factor affecting the generator performance. At the same time, the effects of harmonic currents on the torque ripple, average torque and eddy current loss of the generator are studied, and the mechanism of the variation of the eddy current loss is also discussed.

• The Transactions of the Korean Institute of Electrical Engineers P
• /
• v.58 no.4
• /
• pp.516-520
• /
• 2009

Atmosphere Plasmas of Gas Discharge (APGD) have been used in plasma sources for material processing such as etching, deposition, surface modification, etc. This study is to investigate and understand the fundamental plasma discharge properties. Especially, $SF_6/N_2$ mixed gas would be used in power transformer, GIS (Gas insulated switchgear) and so on. In this paper, we developed a one dimensional fluid simulation model with capacitively coupled plasma chamber at the atmosphere pressure (760 [Torr]). 38 kinds of $SF_6/N_2$ plasma particles which are an electron, two positive ions (${SF_5}^+$, ${N_2}^+$), five negative ions (${SF_6}^-$, ${SF_5}^-$, ${SF_4}^-$, ${F_2}^-$, ${F_1}^-$), thirty excitation and vibrational particles for $N_2$ were considered in this computation. The $N_2$ gases of 20%, 50%, 80% were mixed in $SF_6$ gas. As the amount of $N_2$ gas was increased, the properties of electro-negative plasma moved toward the electro-positive plasma.

• Proceedings of the KIEE Conference
• /
• 2000.07a
• /
• pp.259-261
• /
• 2000

UPFC(Unified Power Flow Controller) consists of two voltage sourced converter(VSC)s inserted into AC system through series and parallel coupling transformer, where two VSCs are linked by capacitor at DC-side. Since VSC acts as an AC voltage source behind a reactance, where both magnitude and phase angle of the source are controllable, UPFC can be represented by the equation related to input-output relation of two VSCs. Voltage control of DC-link capacitor provides the path of real power flow between two VSCs. While UPFC is controlled for maintaining the given reference value in steady state, it should be controlled for damping power oscillation in dynamics. For such a control objective, the control strategy based on the energy function was proposed and has been shown to be effect and robust for damping power oscillation of power system. In this paper, UPFC model based on the VSC was analysed and applied to power-flow control and stability analysis. The control strategy based on the energy function is adopted for damping power oscillation of power system. The effectiveness of proposed control strategy was verified by simulation study

• Proceedings of the KIEE Conference
• /
• 2005.07a
• /
• pp.96-98
• /
• 2005

In this paper, an algorithm to compensate the distorted signals due to Current Transformer(CT) saturation is suggested, First, DWT which can be easily realized by filter banks in real-time applications is used to detect a start point and an end point of the saturation. Secondly, For enough Datas those need to use the least-square curve fitting method, the distorted current signal is compensated by the AR(autoregressive) model using the data during the previous healthy section until pick point of Saturation. Thirdly, the least-square curve fitting method is used to restore the distorted section of the secondary current. Finaly, this algorithm had a Hadware test using DSP board(TMS320C32) with Doble test device. DWT has superior detection accuracy and the proposed compensation algorithm which shows very stable features under various levels of remanent flux in the CT core is also satisfactory. And this algorithm is more correct than a previous algorithm which is only using the LSQ fitting method. Also it can be used as a MU involving the compensation function that acquires the second data from CT and PT.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.15 no.10 s.89
• /
• pp.1005-1010
• /
• 2004

An analysis method of slot-coupled microstripline to waveguide transition is presented to developed a simple but accurate equivalent circuit model. The equivalent circuit consists of an ideal transformer, microstrip open stub, and admittance elements looking into a waveguide and a half space of feed side from a slot center. The related circuit element values are calculated by applying the reciprocity theorem, the Fourier transform and series representation, the complex power concept, and the spectral-domain immittance approach. The computed scattering parameters are compared with the measured, and good agreement validates the simplicity and accuracy of the proposed equivalent circuit model.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.12 no.9
• /
• pp.3795-3801
• /
• 2011

The mathematical simulation of voltage and current waveform of the discharge lamp is useful for the analysis and design of ballasting circuits. This paper proposes a mathematical model which has lamp power and negative voltage drop in discharge lamp. Simulation applying the proposed model has been done, and the results are compared with the experimental results. Furthermore, in the paper, the ballast components(core, transformer) was designed such that high intensity discharge could be optimized for the automotive, by applying a method simulation based design.