• 한국전자통신학회논문지
• /
• 제5권3호
• /
• pp.315-320
• /
• 2010

다중 클래스로 이루어진 패턴을 분류하는 것은 데이터 베이스 시스템에서 기록을 그룹화하거나 VLSI 회로에서 어디에 결함이 있는지를 찾는 것 등에서 중요한 역할을 한다. 본 논문에서는 주어진 다중 클래스 패턴을 MACA(Multiple Attractor Cellular Automata)와 부분공간의 개념을 이용하여 가능한 최소 메모리량을 필요로 하는 다중 클래스 패턴 분류기를 구성하는 알고리즘을 제안한다.

• 대한원격탐사학회지
• /
• 제22권3호
• /
• pp.211-219
• /
• 2006

A random forests classifier is applied to multi-sensor data fusion for supervised land-cover classification in order to account for the importance of variable. The random forests approach is a non-parametric ensemble classifier based on CART-like trees. The distinguished feature is that the importance of variable can be estimated by randomly permuting the variable of interest in all the out-of-bag samples for each classifier. Two different multi-sensor data sets for supervised classification were used to illustrate the applicability of random forests: one with optical and polarimetric SAR data and the other with multi-temporal Radarsat-l and ENVISAT ASAR data sets. From the experimental results, the random forests approach could extract important variables or bands for land-cover discrimination and showed reasonably good performance in terms of classification accuracy.

• 스마트미디어저널
• /
• 제13권2호
• /
• pp.9-15
• /
• 2024

본 논문에서는 신용 예측에서 발생하는 불균형 문제를 해결하기 위해 Diffusion Multi-step Classifier(DMC)를 제안한다. DMC는 Diffusion Model을 통해 신용 예측 데이터의 연속적인 수치형 데이터들을 생성하고 생성된 데이터들을 Multi-step Classifier로 구분하는 것으로 범주형 데이터를 생성한다. DMC를 통해 기존의 데이터를 생성하는 다른 알고리즘보다 실제 데이터와 유사한 분포를 가지는 데이터를 생성할 수 있었다. 이렇게 생성된 데이터를 사용하여 실험을 진행하였을 때 연체를 예측할 확률이 20%이상 상승하였으며, 전체적으로 예측 정확성은 약 4%정도 상승하였다. 이러한 연구 결과는 실제 금융기관에 적용 시 연체율 감소와 수익 증가에 큰 기여를 할 수 있을것으로 예상된다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2002년도 ICCAS
• /
• pp.109.3-109
• /
• 2002

1. Introduction 2. Vision system 3. Effect of brightness variations 4. Color classifier using multi-layer neural network 5. Experimental result of color classifier 6. Applications for robot soccer system 7. Conclusion

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제9권12호
• /
• pp.5087-5102
• /
• 2015

Rapid popularization of smart cell phone equipped with camera has led to a number of new legal and criminal problems related to multimedia such as digital image, which makes cell phone source identification an important branch of digital image forensics. This paper proposes a classifier combination based source identification strategy for cell phone images. To identify the outlier cell phone models of the training sets in multi-class classifier, a one-class classifier is orderly used in the framework. Feature vectors including color filter array (CFA) interpolation coefficients estimation and multi-feature fusion is employed to verify the effectiveness of the classifier combination strategy. Experimental results demonstrate that for different feature sets, our method presents high accuracy of source identification both for the cell phone in the training sets and the outliers.

• 대한원격탐사학회지
• /
• 제37권4호
• /
• pp.719-731
• /
• 2021

This study proposes a two-stage hybrid classification model for crop classification using multi-temporal remote sensing images; the model combines feature embedding by using an autoencoder (AE) with a convolutional neural network (CNN) classifier to fully utilize features including informative temporal and spatial signatures. Long short-term memory (LSTM)-based AE (LAE) is fine-tuned using class label information to extract latent features that contain less noise and useful temporal signatures. The CNN classifier is then applied to effectively account for the spatial characteristics of the extracted latent features. A crop classification experiment with multi-temporal unmanned aerial vehicle images is conducted to illustrate the potential application of the proposed hybrid model. The classification performance of the proposed model is compared with various combinations of conventional deep learning models (CNN, LSTM, and convolutional LSTM) and different inputs (original multi-temporal images and features from stacked AE). From the crop classification experiment, the best classification accuracy was achieved by the proposed model that utilized the latent features by fine-tuned LAE as input for the CNN classifier. The latent features that contain useful temporal signatures and are less noisy could increase the class separability between crops with similar spectral signatures, thereby leading to superior classification accuracy. The experimental results demonstrate the importance of effective feature extraction and the potential of the proposed classification model for crop classification using multi-temporal remote sensing images.

• 전기학회논문지
• /
• 제67권11호
• /
• pp.1506-1511
• /
• 2018

This study presents a human sensibility evaluation method using neural network and multiple-template method on electroencephalogram(EEG). We used a multi-layer perceptron type neural network as the sensibility classifier using EEG signal. For our research objective, 10-channel EEG signals are collected from the healthy subjects. After the necessary preprocessing is performed on the acquired signals, the various EEG parameters are estimated and their discriminating performance is evaluated in terms of pattern classification capability. In our study, Linear Prediction(LP) coefficients are utilized as the feature parameters extracting the characteristics of EEG signal, and a multi-layer neural network is used for indicating the degree of human sensibility. Also, the estimation for human comfortableness is performed by varying temperature and humidity environment factors and our results showed that the proposed scheme achieved good performances for evaluation of human sensibility.

• 정보보호학회논문지
• /
• 제28권3호
• /
• pp.617-623
• /
• 2018

본 연구는 안드로이드 정적분석을 기반으로 추출된 AndroidManifest 권한 특징을 통해 악성코드를 탐지하고자 한다. 특징들은 AndroidManifest의 권한을 기반으로 분석에 대한 자원과 시간을 줄였다. 악성코드 탐지 모델은 1500개의 정상어플리케이션과 500개의 악성코드들을 학습한 SVM(support vector machine), NB(Naive Bayes), GBC(Gradient Boosting Classifier), Logistic Regression 모델로 구성하여 98%의 탐지율을 기록했다. 또한, 악성앱 패밀리 식별은 알고리즘 SVM과 GPC (Gaussian Process Classifier), GBC를 이용하여 multi-classifiers모델을 구현하였다. 학습된 패밀리 식별 머신러닝 모델은 악성코드패밀리를 92% 분류했다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2003년도 학술회의 논문집 정보 및 제어부문 B
• /
• pp.588-591
• /
• 2003

This paper presents a new short-term load forecasting system using data mining. Since the electric load has very different pattern according to the day, it definitely gives rise to the forecasting error if only one forecasting model is used. Thus, to resolve this problem, the fuzzy model-based classifier and predictor are proposed for the forecasting of the hourly electric load. The proposed classifier is the multi-input and multi-output fuzzy system of which the consequent part is composed of the Bayesian classifier. The proposed classifier attempts to categorize the input electric load into Monday, Tuesday$\sim$Friday, Saturday, and Sunday electric load, Then, we construct the Takagi-Sugeno (T-S) fuzzy model-based predictor for each class. The parameter identification problem is converted into the generalized eigenvalue problem (GEVP) by formulating the linear matrix inequalities (LMIs). Finally, to show the feasibility of the proposed method, this paper provides the short-term load forecasting example.

• Journal of Information Processing Systems
• /
• 제14권5호
• /
• pp.1114-1135
• /
• 2018

A multiple classification system based on a new boosting technique has been approached utilizing different biometric traits, that is, color face, iris and eye along with fingerprints of right and left hands, handwriting, palm-print, gait (silhouettes) and wrist-vein for person authentication. The images of different biometric traits were taken from different standard databases such as FEI, UTIRIS, CASIA, IAM and CIE. This system is comprised of three different super-classifiers to individually perform person identification. The individual classifiers corresponding to each super-classifier in their turn identify different biometric features and their conclusions are integrated together in their respective super-classifiers. The decisions from individual super-classifiers are integrated together through a mega-super-classifier to perform the final conclusion using programming based boosting. The mega-super-classifier system using different super-classifiers in a compact form is more reliable than single classifier or even single super-classifier system. The system has been evaluated with accuracy, precision, recall and F-score metrics through holdout method and confusion matrix for each of the single classifiers, super-classifiers and finally the mega-super-classifier. The different performance evaluations are appreciable. Also the learning and the recognition time is fairly reasonable. Thereby making the system is efficient and effective.