• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.34 no.1
• /
• pp.11-18
• /
• 2022

In order to improve the accuracy of particle tracking prediction techniques near the Korean Strait, this study compared and analyzed a particle tracking model based on a seawater flow numerical model and a machine learning based on a particle tracking model using field observation data. The data used in the study were the surface drifter buoy movement trajectory data observed in the Korea Strait, prediction data by machine learning (linear regression, decision tree) using the tide and wind data from three observation stations (Gageo Island, Geoje Island, Gyoboncho), and prediciton data by numerical models (ROMS, MOHID). The above three data were compared through three error evaluation methods (Correlation Coefficient (CC), Root Mean Square Errors (RMSE), and Normalized Cumulative Lagrangian Separation (NCLS)). As a final result, the decision tree model had the best prediction accuracy in CC and RMSE, and the MOHID model had the best prediction results in NCLS.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.34 no.3
• /
• pp.72-81
• /
• 2022

Positions of five drifting buoys deployed on August 2020 near southwestern area of Jeju Island and numerically predicted velocities were used to develop five Artificial Intelligence-based models (AI models) for the prediction of particle tracks. Five AI models consisted of three machine learning models (Extra Trees, LightGBM, and Support Vector Machine) and two deep learning models (DNN and RBFN). To evaluate the prediction accuracy for six models, the predicted positions from five AI models and one numerical model were compared with the observed positions from five drifting buoys. Three skills (MAE, RMSE, and NCLS) for the five buoys and their averaged values were calculated. DNN model showed the best prediction accuracy in MAE, RMSE, and NCLS.

• Korean Journal of Remote Sensing
• /
• v.39 no.6_2
• /
• pp.1577-1589
• /
• 2023

Recently, the necessity of predicting changes for monitoring ocean is widely recognized. In this study, we performed a time series prediction of remote-sensing reflectance (Rrs), which can indicate changes in the ocean, using Geostationary Ocean Color Imager (GOCI) data. Using GOCI-I data, we trained a multi-scale Convolutional Long-Short-Term-Memory (ConvLSTM) which is proposed in this study. Validation was conducted using GOCI-II data acquired at different periods from GOCI-I. We compared model performance with the existing ConvLSTM models. The results showed that the proposed model, which considers both spatial and temporal features, outperformed other models in predicting temporal trends of Rrs. We checked the temporal trends of Rrs learned by the model through long-term prediction results. Consequently, we anticipate that it would be available in periodic change detection.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 2020.11a
• /
• pp.177-178
• /
• 2020

본 논문은 선박항해 에이전트가 개발 의도와 부합되도록 학습시키는데 있어 가장 중요한 역할을 수행하는 보상설계에 대해 소개한다. 보상설계는 먼저 학습 대상이 무엇인지 명확히 정의하는 것이 중요하며, 보상이 상황에 따라 다른 목적으로 활용되지 않도록 하고 에이전트에게 너무 드물게 주어지지 않도록 보상 형태화를 적용하는 등의 방법을 사용할 필요가 있다. 또한 보상을 구성하는 요소가 많아지는 경우에는 의도가 명확하게 전달이 되지 않을 수 있으므로 문제를 작은 문제들로 나누어 접근하는 계층적 강화학습 방법 등을 적용할 필요가 있다.

• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.179-183
• /
• 2020

말뭉치는 기계학습 및 심층학습을 위한 필수 자원이다. 한국어 개체명의 경우 학습에 사용할 잘 정제된 개체명 부착 말뭉치가 충분하지 않다. 말뭉치 정제 작업은 시간적, 경제적으로 많은 비용이 소모된다. 따라서 본 논문에서는 적은 양의 말뭉치를 이용하여 말뭉치를 자동적으로 확장하는 방법을 제안한다. 특별히 소규모 말뭉치에 속하는 문장의 단어에 대한 형제어들을 선정하여 형제어의 확률추출을 기반으로 대체함으로써 새로운 문장을 생성함으로써 말뭉치 확장하는 방법이다. 본 논문에서는 확장된 말뭉치를 이용해서 대부분의 시스템에서 성능이 향상됨을 확인할 수 있었다. 앞으로 단어의 삭제 및 삽입 등 다양한 방법으로 좀 더 다양한 문장을 생성할 수 있을 것으로 생각합니다.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.6
• /
• pp.529-535
• /
• 2023

Noise originating from incipient propeller cavitation is assumed to come from a limited number of sources emitting a broadband signal. Conventional methods for cavitation localization have limitations because they cannot distinguish adjacent sound sources effectively due to low accuracy and resolution. On the other hand, sparse Bayesian learning technique demonstrates high-resolution restoration performance for sparse signals and offers greater resolution compared to conventional cavitation localization methods. In this paper, an incipient propeller cavitation localization method using sparse Bayesian learning is proposed and shown to be superior to the conventional method in terms of accuracy and resolution through experimental data from a model ship.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.32 no.4
• /
• pp.331-339
• /
• 1996

It is most important to understand the behaviour of fish in case of the betterment of the current fishing gear and methods or the development of the conditioning by acoustic sound in marine ranching. This investigation has been attempted to provide for the prediction of the response action of fish to acoustic sound. The experimental fish was conditioned with sound and bait. As the acoustic sound for stimulus, the pure tone of sine waveform at the frequency of 200Hz was used. This pure tone was determined from previous investigation about hearing ability of walleye pollock Theragra chalcogramma. The fork length of walleye pollock used in this experiment was 385~450mm. The conditioning proceeding was recorded in the video tape recorder. Frequency of appearance in the feeding area was analyzed with computer and video tape recorder. The position of fish was tracked using the mouse cursor and picture mixed on the superimpose board. The response of conditioned fish to sound stimulus was appeared in the 8th day firstly. The conditioned fish remembered the stimulus sound for 4 days. Average frequency of appearance in the feeding area during the 30 seconds sound projection or 1 minute after the sound stimulus was 51%, and was higher than before it.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.634-636
• /
• 2020

본 논문에서는 딥러닝 (deep learning) 방식을 이용한 5G NR (fifth-generation new radio)의 cell ID (cell identity) 검출 기법을 구현하였다. 5G NR 시스템의 단말 (user equipment)은 초기 접속 (initial access)과정에서 PSS (primary synchronization signal)와 SSS (secondary synchronization signal)을 이용한 동기 획득 및 cell ID 검출이 필요하다. 본 논문에서는 분류 기법 기반의 딥러닝 기술을 이용하여 인공 신경망 모델에 PSS 및 SSS 와 cell ID 의 상관 관계를 학습시키고, 학습된 모델의 성능을 제시하였다.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.630-633
• /
• 2022

속성기반 감성 분석은 텍스트 내에 존재하는 속성에 대해 세분화된 감성 분석을 수행하는 과제를 말한다. 세분화된 감성분석을 정확하게 수행하기 위해서는 텍스트에 존재하는 감성 표현과 그것이 수식하는 대상에 대한 정보가 반드시 필요하다. 그리고 순서대로 두 가지 정보는 이후 정보를 텍스트에서 추출하기 위해 중요한 단서가 된다. 따라서 본 논문에서는 KorBERT와 Bi-LSTM을 이용한 단계적 특징을 활용한 다중 작업 학습 모델을 사용하여 한국어 감성 분석 말뭉치의 감성 표현과 대상을 추출하는 작업을 수행하였다. 제안한 모델을 한국어 감성 분석 말뭉치로 학습 및 평가한 결과, 감성 표현 추출 작업의 출력을 추가적인 특성으로 전달하여 대상 추출 작업의 성능을 향상시킬 수 있음을 보였다.

• Proceedings of KOSOMES biannual meeting
• /
• 2018.06a
• /
• pp.204-204
• /
• 2018