• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39C no.10
• /
• pp.887-895
• /
• 2014

Drowsy driving is a large proportion of the total car accidents. For this reason, drowsiness detection and warning system for drivers has recently become a very important issue. Monitoring physiological signals provides the possibility of detecting features of drowsiness and fatigue of drivers. Many researches have been published that to measure electroencephalogram(EEG) signals is the effective way in order to be aware of fatigue and drowsiness of drivers. The aim of this study is to extract drowsiness-related features from a set of EEG signals and to classify the features into three states: alertness, transition, and drowsiness. This paper proposes a drowsiness detection system using errors-in-variables(EIV) for extraction of feature vectors and multilayer perceptron (MLP) for classification. The proposed method evaluates robustness for noise and compares to the previous one using linear predictive coding (LPC) combined with MLP. From evaluation results, we conclude that the proposed scheme outperforms the previous one in the low signal-to-noise ratio regime.

• KIPS Transactions on Computer and Communication Systems
• /
• v.9 no.11
• /
• pp.247-249
• /
• 2020

In the era of the 4th industrial revolution, the demand for convergence between ICT technologies is increasing in various fields. Accordingly, a new term that combines data, network, and artificial intelligence technology, DNA (Data, Network, AI) is in use. and has recently become a hot topic. DNA has various potential technology to be able to develop intelligent application in the real world. Therefore, this paper introduces the reviewed papers on the service image placement mechanism based on the logical fog network, the mobility support scheme based on machine learning for Industrial wireless sensor network, the prediction of the following BCI performance by means of spectral EEG characteristics, the warning classification method based on artificial neural network using topics of source code and natural language processing model for data visualization interaction with chatbot, related on DNA technology.

• The Journal of the Acoustical Society of Korea
• /
• v.14 no.4
• /
• pp.46-63
• /
• 1995

This thesis described an unlimited vocabulary connected speech recognition system using Time Delay Neural Network(TDNN). The recognition unit is the diphone unit which includes the transition section of two phonemes, and the number of diphone unit is 329. The recognition processing of korean connected speech is composed by three part; the feature extraction section of the input speech signal, the diphone recognition processing and post-processing. In the feature extraction section, the extraction of diphone interval in input speech signal is carried and then the feature vectors of 16th filter-bank coefficients are calculated for each frame in the diphone interval. The diphone recognition processing is comprised by the three stage hierachical structure and is carried using 30 Time Delay Neural Networks. particularly, the structure of TDNN is changed so as to increase the recognition rate. The post-processing section, mis-recognized diphone strings are corrected using the probability of phoneme transition and the probability o phoneme confusion and then the eojeols (Korean word or phrase) are formed by combining the recognized diphones.

• Journal of Korea Water Resources Association
• /
• v.37 no.6
• /
• pp.449-460
• /
• 2004

The purpose of this study is to test the applicability of neuro-fuzzy system for monthly dam inflow forecasts by using weather forecasting information. The neuro-fuzzy algorithm adopted in this study is the ANFIS(Adaptive neuro-fuzzy Inference System) in which neural network theory is combined with fuzzy theory. The ANFIS model can experience the difficulties in selection of a control rule by a space partition because the number of control value increases rapidly as the number of fuzzy variable increases. In an effort to overcome this drawback, this study used the subtractive clustering which is one of fuzzy clustering methods. Also, this study proposed a method for converting qualitative weather forecasting information to quantitative one. ANFIS for monthly dam inflow forecasts was tested in cases of with or without weather forecasting information. It can be seen that the model performances obtained from the use of past observed data and future weather forecasting information are much better than those from past observed data only.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.7C
• /
• pp.554-564
• /
• 2012

The vehicle recognition consists of two steps; the vehicle region detection step and the vehicle identification step based on the feature extracted from the detected region. Features using linear transformations have the effect of dimension reduction as well as represent statistical characteristics, and show the robustness in translation and rotation of objects. Among the linear transformations, the NMF(Non-negative Matrix Factorization) is one of part-based representation. Therefore, we can extract NMF features with sparsity and improve the vehicle recognition rate by the representation of local features of a car as a basis vector. In this paper, we propose a feature extraction using NMF suitable for the vehicle recognition, and verify the recognition rate with it. Also, we compared the vehicle recognition rate for the occluded area using the SNMF(sparse NMF) which has basis vectors with constraint and LVQ2 neural network. We showed that the feature through the proposed NMF is robust in the urban scene where occlusions are frequently occur.

• Journal of Digital Convergence
• /
• v.13 no.3
• /
• pp.155-165
• /
• 2015

Bankruptcy prediction has been one of the important research topics in finance since 1960s. In Korea, it has gotten attention from researchers since IMF crisis in 1998. This study aims at proposing a novel model for better bankruptcy prediction by converging three techniques - support vector machine(SVM), fuzzy theory, and genetic algorithm(GA). Our convergence model is basically based on SVM, a classification algorithm enables to predict accurately and to avoid overfitting. It also incorporates fuzzy theory to extend the dimensions of the input variables, and GA to optimize the controlling parameters and feature subset selection. To validate the usefulness of the proposed model, we applied it to H Bank's non-external auditing companies' data. We also experimented six comparative models to validate the superiority of the proposed model. As a result, our model was found to show the best prediction accuracy among the models. Our study is expected to contribute to the relevant literature and practitioners on bankruptcy prediction.

• Journal of Digital Convergence
• /
• v.14 no.8
• /
• pp.233-243
• /
• 2016

Due to speech recognition problems in noisy environment, Audio Visual Speech Recognition (AVSR) system, which combines speech information and visual information, has been proposed since the mid-1990s,. and lip reading have played significant role in the AVSR System. This study aims to enhance recognition rate of utterance word using only lip shape detection for efficient AVSR system. After preprocessing for lip region detection, Convolution Neural Network (CNN) techniques are applied for utterance period detection and lip shape feature vector extraction, and Hidden Markov Models (HMMs) are then used for the recognition. As a result, the utterance period detection results show 91% of success rates, which are higher performance than general threshold methods. In the lip reading recognition, while user-dependent experiment records 88.5%, user-independent experiment shows 80.2% of recognition rates, which are improved results compared to the previous studies.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.19 no.5
• /
• pp.187-194
• /
• 2019

Traditional demand forecasting methods are difficult to meet the needs of companies due to rapid changes in the market and the diversification of individual consumer needs. In a diversified production environment, the right demand forecast is an important factor for smooth yield management. Many of the existing predictive models commonly used in industry today are limited in function by little. The proposed model is designed to overcome these limitations, taking into account the part where each model performs better individually. In this paper, variables are extracted through Gray Relational analysis suitable for dynamic process analysis, and statistically predicted data is generated that includes characteristics of historical demand data produced through ARIMA forecasts. In combination with the LSTM model, demand forecasts can then be calculated by reflecting the many factors that affect demand forecast through an architecture that is structured to avoid the long-term dependency problems that the neural network model has.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2018.05a
• /
• pp.6-6
• /
• 2018

최근 집중 호우로 인한 내수침수 피해가 도시화와 기후변화로 늘어나고 있다. 내수침수 피해로 인한 복구비용과 시간이 증가하고 있으며 향후에는 이보다 더 크게 늘어날 것으로 예상된다. 이러한 문제를 해결하기 위하여 충분한 선행시간을 가지고 내수 침수 구역을 제시할 수 있어야 한다. 기존의 물리적 모델은 정확하고 정교한 결과를 제공하지만, 시뮬레이션을 준비하고 마치는 데에 시간이 많이 소요된다. 그 이유로서는 강우량, 지형적 특성, 배수관망 시스템, 수문학적 매개변수 등의 다양한 데이터도 필요하기 때문이다. 이는 도시유역에 대한 내수침수의 실시간 예측이 어렵게 되었으며, 충분한 선행시간을 확보하지 못하는 원인이 되었다. 본 연구에서는 이 문제에 대한 해결책으로 결정론적 방법과 확률론적 방법을 자료지향형 모형으로 결합하여 해결책을 제시하고자 하며, 특정 강우 조건하에 도시유역에서의 내수침수에 영향을 미치는 맨홀에 대한 정보를 제공하고자 한다. 위와 같은 과정을 수행하기 위하여 입력자료 조합에 대한 비선형 분석을 실시하였으며, 그 결과로 특정 강우 조건에 대하여 각 맨홀에 대한 누적월류량을 예측할 수 있는 비선형 인공신경망을 구축할 수 있었다. 본 연구에서 제시된 방법론은 국내의 강남 배수분구에 대하여 적용이 되었으며, 내수침수 예측결과와 2차원 해석결과를 비교하고자 하였다. 본 연구에서는 위 과정을 통하여 1차원 도시유출해석을 위한 입력 자료를 준비하는 시간을 절약하고, 다양한 강우 조건과 내수침수지도 사이의 연관성을 학습하는 예측 모형을 이용하여 도시유역의 내수침수에 대한 충분한 선행시간을 확보하고자 한다. 결론적으로, 이 연구의 결과는 도시유역에 대한 비구조적 대책 수립에 도움을 줄 것으로 확인이 되며 도시 유역 내에 맨홀 위치들을 고려한 위험지구를 파악하는 데에 유용할 것으로 판단된다.

• Phonetics and Speech Sciences
• /
• v.13 no.3
• /
• pp.71-78
• /
• 2021

Most neural-network-based speech synthesis models utilize neural vocoders to convert mel-scaled spectrograms into high-quality, human-like voices. However, neural vocoders combined with mel-scaled spectrogram prediction models demand considerable computer memory and time during the training phase and are subject to slow inference speeds in an environment where GPU is not used. This problem does not arise in linear spectrogram prediction models, as they do not use neural vocoders, but these models suffer from low voice quality. As a solution, this paper proposes a Tacotron 2 and Transformer-based linear spectrogram prediction model that produces high-quality speech and does not use neural vocoders. Experiments suggest that this model can serve as the foundation of a high-quality text-to-speech model with fast inference speed.