• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.2
• /
• pp.163-168
• /
• 2003

Several equipments for flame detection are employed in the power generations. But these flame detectors have some problems for the correct performance. So in this paper, we apply different techniques for the flame detection. Image processing techniques are broadly applied in industrial fields. In this paper, the image information is recorded by a camcoder and then these images are preprocessed for the input values of neural network model. We can test and evaluate the approach that uses image information for the flame detection of burners. If this technique is implemented in physical plant, the economical and effective operation could be achieved.

• Journal of Korea Water Resources Association
• /
• v.32 no.1
• /
• pp.3-13
• /
• 1999

In this study, an alternative spatial analysis method against conventional methods such as Thiessen method, Inverse Distance method, and Kriging method, named Spatial-Analysis Neural-Network (SANN) is presented. It is based on neural network modeling and provides a nonparametric mean estimator and also estimators of high order statistics such as standard deviation and skewness. In addition, it provides a decision-making tool including an estimator of posterior probability that a spatial variable at a given point will belong to various classes representing the severity of the problem of interest and a Bayesian classifier to define the boundaries of subregions belonging to the classes. In this paper, the SANN is implemented to be used for analyzing a mean annual precipitation filed and classifying the field into dry, normal, and wet subregions. For an example, the whole area of South Korea with 39 precipitation sites is applied. Then, several useful results related with the spatial variability of mean annual precipitation on South Korea were obtained such as interpolated field, standard deviation field, and probability maps. In addition, the whole South Korea was classified with dry, normal, and wet regions.

• The Journal of the Acoustical Society of Korea
• /
• v.37 no.4
• /
• pp.256-261
• /
• 2018

There are three types of noise generated inside the vehicle: BSR (Buzz, Squeak, Rattle). In this paper, we propose a classifier that automatically classifies automotive BSR noise by using features extracted from deep convolutional neural networks. In the preprocessing process, the features of above three noises are represented as noise-map using STFT (Short-time Fourier Transform) algorithm. In order to cope with the problem that the position of the actual noise is unknown in the part of the generated noise map, the noise map is divided using the sliding window method. In this paper, internal parameter of the deep convolutional neural networks is visualized using the t-SNE (t-Stochastic Neighbor Embedding) algorithm, and the misclassified data is analyzed in a qualitative way. In order to analyze the classified data, the similarity of the noise type was quantified by SSIM (Structural Similarity Index) value, and it was found that the retractor tremble sound is most similar to the normal travel sound. The classifier of the proposed method compared with other classifiers of machine learning method recorded the highest classification accuracy (99.15 %).

• The Journal of the Korea Contents Association
• /
• v.11 no.4
• /
• pp.27-32
• /
• 2011

This paper presents a methodology for classifying sleep disturbance using electroencephalogram (EEG) signal at sleep stage 2 and principal component analysis. For extracting initial features, fast Fourier transforms(FFT) were carried out to remove some noise from EEG signal at sleep stage 2. In the second phase, we used principal component analysis to reduction from EEG signal that was removed some noise by FFT to 5 features. In the final phase, 5 features were used as inputs of NEWFM to get performance results. The proposed methodology shows that accuracy rate, specificity rate, and sensitivity were all 100%.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.312-318
• /
• 2022

This paper proposes a model which classifies the type of guns and information about sound source location using deep neural network. The proposed classification model is composed of convolutional neural networks (CNN) and long short-term memory (LSTM). For training and test the model, we use the Gunshot Audio Forensic Dataset generated by the project supported by the National Institute of Justice (NIJ). The acoustic signals are transformed to Mel-Spectrogram and they are provided as learning and test data for the proposed model. The model is compared with the control model consisting of convolutional neural networks only. The proposed model shows high accuracy more than 90 %.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.06b
• /
• pp.277-279
• /
• 2006

본 논문에서는 스크래치 텍스처 및 형태특성을 이용하여 모든 종류의 스크래치를 자동으로 검출 할 수 있는 방법을 제안한다. 제안한 방법은 텍스처 분류 단계와 형태 필터링 단계를 구성된다. 텍스처 분류단계에서는 스크래치의 텍스처 정보를 이용하여 입력영상의 각 화소를 스크래치와 비스크래래치 영역으로 분류한다. 이때 분류기로 신경망을 사용한다. 형태필터링단계에서는 스크래치의 형태정보에 기반하여 설계된 원소구조를 사용하는 모폴로지 필터를 사용하여 잘못 분류된 스크래치 영역을 제거한다. 제안된 방법의 평가를 위해 다양한 종류의 스크래치를 가진 영화 및 애니메이션 데이터에 대해 실험이 이루어 졌고, 그 결과 제안된 방법의 강건함과 효율성이 입증되었다.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.103-108
• /
• 2019

한국어 질의 응답의 입력 질문에 대한 예상 정답 유형을 단답형 또는 서술형으로 이진 분류하는 방법에 대해 서술한다. 일반적인 개체명 인식으로 확인할 수 없는 질의 주제어의 화제성을 반영하기 위하여, 검색 엔진 쿼리를 빈도수로 분석한다. 분석된 질의 주제어 정보와 함께, 정답의 범위를 제약할 수 있는 속성 표현과 육하원칙 정보를 입력 자질로 사용한다. 기존 신경망 분류 모델과 비교한 실험에서, 추가 자질을 적용한 모델이 4% 정도 향상된 분류 성능을 보이는 것을 확인할 수 있었다.

• Proceedings of the KIEE Conference
• /
• 2003.07d
• /
• pp.2552-2554
• /
• 2003

난수발생기 (Random generator-RG)와 GRNN을 이용한 분류기 설계방식을 제안하며, 이를 프로모터 염기서열의 분류에 적용한다. 주어진 난수범위에서 다중 분류기를 발생하였으며, 그 성능을 예측정확도와 분류민감도 측면에서 평가하였고, 분류민감도는 다시 전체와 개별적 프로모터에 대해서 세분화하여 평가하였다. 최적화된 분류기 상호간의 비교에서 제안된 기법은 모든 임계점에 대해서, 전체 분류민감도와 전체 예측정확도를 향상시키었으며, 이는 전체 분류 민감도에서 더 두드러졌다. 한편, 개별적 프로모터에 대한 분류민감도와 예측정확도도 평균적으로 향상되었다. 이 같은 결과로 제안된 기법이 분류와 예측성능을 동시에 증진하는데 매우 효과적임을 알 수 있었다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.22 no.3
• /
• pp.319-327
• /
• 2012

In this paper, we present a sign language recognition model which does not use any wearable devices for object tracking. The system design issues and implementation issues such as data representation, feature extraction and pattern classification methods are discussed. The proposed data representation method for sign language patterns is robust for spatio-temporal variances of feature points. We present a feature extraction technique which can improve the computation speed by reducing the amount of feature data. A neural network model which is capable of incremental learning is described and the behaviors and learning algorithm of the model are introduced. We have defined a measure which reflects the relevance between the feature values and the pattern classes. The measure makes it possible to select more effective features without any degradation of performance. Through the experiments using six types of sign language patterns, the proposed model is evaluated empirically.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.6
• /
• pp.629-636
• /
• 2022

We present a preprocessing method for an underwater target detection model based on a convolutional neural network. The acoustic characteristics of the ship show ambiguous expression due to the strong signal power of the low frequency. To solve this problem, we combine feature preprocessing methods with various feature scaling methods and spectrogram methods. Define a simple convolutional neural network model and train it to measure preprocessing performance. Through experiment, we found that the combination of log Mel-spectrogram and standardization and robust scaling methods gave the best classification performance.