• 한국정보시스템학회지:정보시스템연구
• /
• 제29권1호
• /
• pp.241-265
• /
• 2020

Purpose Various machine learning techniques are used to implement for predicting corporate credit. However, previous research doesn't utilize time series input features and has a limited prediction timing. Furthermore, in the case of corporate bond credit rating forecast, corporate sample is limited because only large companies are selected for corporate bond credit rating. To address limitations of prior research, this study attempts to implement a predictive model with more sample companies, which can adjust the forecasting point at the present time by using the credit score information and corporate information in time series. Design/methodology/approach To implement this forecasting model, this study uses the sample of 2,191 companies with KIS credit scores for 18 years from 2000 to 2017. For improving the performance of the predictive model, various financial and non-financial features are applied as input variables in a time series through a sliding window technique. In addition, this research also tests various machine learning techniques that were traditionally used to increase the validity of analysis results, and the deep learning technique that is being actively researched of late. Findings RNN-based stateful LSTM model shows good performance in credit rating prediction. By extending the forecasting time point, we find how the performance of the predictive model changes over time and evaluate the feature groups in the short and long terms. In comparison with other studies, the results of 5 classification prediction through label reclassification show good performance relatively. In addition, about 90% accuracy is found in the bad credit forecasts.

• 대한원격탐사학회지
• /
• 제30권4호
• /
• pp.493-503
• /
• 2014

이 논문에서는 대규모 작물 재배 지역의 작물 분류도의 조기 제작을 목적으로 분광학적 혼재를 줄이고, 과거 토지피복도의 작물 재배 패턴을 반영할 수 있는 계층적 분류 방법론을 제안하였다. 특히 작물 생육 주기로부터 다른 분광 특성을 고려한 계층적 분류 접근을 적용하고, 과거 작물 재배 패턴으로부터 추출된 시간적 문맥 정보를 함께 고려함으로써 분광 혼재가 두드러진 화소의 영향을 줄일 수 있다. 제안 분류 기법의 적용성을 평가하기 위해 미국 아이오와 주 전체를 대상으로 시계열 MODIS 250 m 정규식생지수 자료와 과거 crop data layer를 사용하는 사례 연구를 수행하였다. 사례 연구를 통해 다른 분류 단계와 과거 작물 재배 패턴을 고려함으로써 대상 지역의 주요 재배 작물이면서 분광학적 유사도가 두드러진 콩과 옥수수를 효과적으로 구분할 수 있었다. 그리고 분광 정보만을 이용한 분류 결과에 비해 제안 기법이 최소 7.68%p에서 최대 20.96%p의 향상된 분류 정확도를 보였다. 또한 분류 단계에서 시간적 문맥 정보를 결합함으로써 사용 NDVI 자료의 수에 영향을 덜 받는 가장 높은 분류 정확도(최대 전체 정확도: 86.63%)를 얻을 수 있었다. 따라서 제안 분류 기법은 주요 곡물 수입국의 대규모 작물 구분도의 조기 제작에 유용하게 사용될 수 있을 것으로 기대된다.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 1991년도 춘계공동학술대회 발표논문 및 초록집; 전북대학교, 전주; 26-27 Apr. 1991
• /
• pp.303-304
• /
• 1991

• 한국지능시스템학회논문지
• /
• 제17권4호
• /
• pp.563-568
• /
• 2007

본 논문은 가중 퍼지소속함수 기반 신경망(Neural Network with Weighted Fuzzy Membership Functions, NEWFM)을 이용하여 클래스의 분류강도를 구하고 비선형 시계열 추이선을 예측하는 방안을 제안하고 있다. NEWFM에 의하여 추출된 가중퍼지 소속함수(BSWFM)를 이용하여 입력값에 대한 분류강도를 구하게 되고, 이들에 대한 가중평균 역퍼지화를 통하여 비선형 시계열 추이선을 작성한다. 실증분석결과 NEWFM은 목표 클래스로 설정된 GDP에 대하여 92.22%의 분류성능을 보여 주었다. 따라서 동 비선형 시계열 추이선은 대표적인 경기지표인 GDP 추이에 비교적 높은 유사도를 나타내는 가운데 분석대상기간인 제5순환기-제8순환기 중 정점(peak)에서 평균 12개월, 저점(trough)에서 평균 6개월의 선행성(look-ahead)을 보여 줌으로써 경기변동에 앞서 상당기간의 시차를 둔 예측지표로서 활용가능성이 입증되었다. NEWFM은 그 특징선택(feature selection)에 의하여 선행지표 10개 중 3개의 축소를 기할 수 있게 해 줌으로써 보다 적은 수의 경제지표를 가지고도 분류성능을 90.0%에서 92.22%로 향상을 기하는 가운데 효율적인 예측기능을 수행할 수 있음이 입증되었다.

• 융합신호처리학회논문지
• /
• 제25권1호
• /
• pp.7-14
• /
• 2024

본 연구는 시계열 데이터의 연속성과 인공지능 모델의 분류 성능에 대한 비교 실험을 수행하였다. EEG 신호를 이용한 BCI 연구에서는 데이터 연속성이 감소할수록 행동과 사고 분류의 성능이 향상되었다. 특히, LSTM은 연속성이 낮은 데이터에서 0.8728이라는 높은 성능을 달성하였고, 연속성을 고려하지 않은 경우 DNN이 0.9178의 성능을 보였다. 연속성을 고려하지 않는 데이터가 더 우수한 성능을 보일 수 있음을 시사하였다. 또한, 연속성을 고려하지 않은 데이터는 작업 분류에서도 더 높은 성능을 보였다. 이러한 결과는 뇌파 신호를 기반으로 한 BCI 연구에서는 데이터 연속성을 고려하기보다는 셔플링을 통해 다양한 데이터 특성을 보여줌으로써 우수한 성능을 발휘할 수 있음을 시사한다.

• 대한원격탐사학회지
• /
• 제40권1호
• /
• pp.57-69
• /
• 2024

Crop classification plays a vitalrole in monitoring agricultural landscapes and enhancing food production. In this study, we explore the effectiveness of Long Short-Term Memory (LSTM) models for crop classification, focusing on distinguishing between apple and rice crops. The aim wasto overcome the challenges associatedwith finding phenology-based classification thresholds by utilizing LSTM to capture the entire Normalized Difference Vegetation Index (NDVI)trend. Our methodology involvestraining the LSTM model using a reference site and applying it to three separate three test sites. Firstly, we generated 25 NDVI imagesfrom the Sentinel-2A data. Aftersegmenting study areas, we calculated the mean NDVI values for each segment. For the reference area, employed a training approach utilizing the NDVI trend line. This trend line served as the basis for training our crop classification model. Following the training phase, we applied the trained model to three separate test sites. The results demonstrated a high overall accuracy of 0.92 and a kappa coefficient of 0.85 for the reference site. The overall accuracies for the test sites were also favorable, ranging from 0.88 to 0.92, indicating successful classification outcomes. We also found that certain phenological metrics can be less effective in crop classification therefore limitations of relying solely on phenological map thresholds and emphasizes the challenges in detecting phenology in real-time, particularly in the early stages of crops. Our study demonstrates the potential of LSTM models in crop classification tasks, showcasing their ability to capture temporal dependencies and analyze timeseriesremote sensing data.While limitations exist in capturing specific phenological events, the integration of alternative approaches holds promise for enhancing classification accuracy. By leveraging advanced techniques and considering the specific challenges of agricultural landscapes, we can continue to refine crop classification models and support agricultural management practices.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권5호
• /
• pp.2509-2528
• /
• 2019

There are various styles of license plates for different countries and use cases that require style-specific methods. In this paper, we propose and illustrate a multi-style license plate recognition system. The proposed system performs a series of processes for license plate candidates detection, structure classification, character segmentation and character recognition, respectively. Specifically, we introduce a license plate structure classification process to identify its style that precedes character segmentation and recognition processes. We use a K-Nearest Neighbors algorithm with pre-training steps to recognize numbers and characters on multi-style license plates. To show feasibility of our multi-style license plate recognition system, we evaluate our system for multi-style license plates covering single line, double line, different backgrounds and character colors on Korean and the U.S. license plates. For the evaluation of Korean license plate recognition, we used a 50 minutes long input video that contains 138 vehicles of 6 different license plate styles, where each frame of the video is processed through a series of license plate recognition processes. From two experiments results, we show that various LP styles can be recognized under 50 ms processing time and with over 99% accuracy, and can be extended through additional learning and training steps.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2003년도 봄 학술발표논문집 Vol.30 No.1 (B)
• /
• pp.178-180
• /
• 2003

EEG-based brain computer interface(BCI) provides a new communication channel between human brain and computer. EEG data is a multivariate time series so that hidden Markov model (HMM) might be a good choice for classification. However EEG is very noisy data and contains artifacts, so useful features mr expected to improve the performance of HMM. In this paper we addresses the usefulness of principal component features with Hidden Markov model (HHM). We show that some selected principal component features can suppress small noises and artifacts, hence improves classification performance. Experimental study for the classification of EEG data during imagination of a left, right up or down hand movement confirms the validity of our proposed method.

• Smart Structures and Systems
• /
• 제26권1호
• /
• pp.23-33
• /
• 2020

Majority of the damage in engineering structures is nonlinear. Damage sensitive features (DSFs) extracted by traditional methods from linear time series models cannot effectively handle nonlinearity induced by structural damage. A new DSF is proposed based on vector space cosine similarity (VSCS), which combines K-means cluster analysis and Bayesian discrimination to detect nonlinear structural damage. A reference autoregressive moving average (ARMA) model is built based on measured acceleration data. This study first considers an existing DSF, residual standard deviation (RSD). The DSF is further advanced using the VSCS, and then the advanced VSCS is classified using K-means cluster analysis and Bayes discriminant analysis, respectively. The performance of the proposed approach is then verified using experimental data from a three-story shear building structure, and compared with the results of existing RSD. It is demonstrated that combining the linear ARMA model and the advanced VSCS, with cluster analysis and Bayes discriminant analysis, respectively, is an effective approach for detection of nonlinear damage. This approach improves the reliability and accuracy of the nonlinear damage detection using the linear model and significantly reduces the computational cost. The results indicate that the proposed approach is potential to be a promising damage detection technique.

• 가족자원경영과 정책
• /
• 제21권3호
• /
• pp.141-159
• /
• 2017

The main purpose of this study is to investigate factors that influence sleep duration and sleep time poverty in terms of family characteristics, child characteristics, and time use. A series of data analyses were conducted on children's time use in two-parent families based on the 2013 Korean Children and Youth Panel Survey. One major finding is that children's sleep duration and the probability of having a sleep time poverty are related to their mothers' job classifications. The factors influencing the duration of sleep time and the sleep time poverty are similar in terms of family characteristics and children's time use. The mother's job classification, family income, number of younger siblings, number of older siblings, children's private tutoring hours, computer game hours, and TV hours are statistically significant factors affecting the duration of sleep time and the probability of having a sleep time poverty. However, the factor with greatest influence on sleep time duration is private tutoring hours and the factor most affecting sleep time poverty is computer game hours. The mother's job classification is a relatively powerful determinant for predicting her children's sleep duration and sleep time poverty.