• Journal of Korea Multimedia Society
• /
• v.24 no.10
• /
• pp.1380-1390
• /
• 2021

The video face recognition (FR) is one of the most popular researches in the field of computer vision due to a variety of applications. In particular, research using the attention mechanism is being actively conducted. In video face recognition, attention represents where to focus on by using the input value of the whole or a specific region, or which frame to focus on when there are many frames. In this paper, we propose a novel attention based deep learning method. Main novelties of our method are (1) the use of combining two loss functions, namely weighted Softmax loss function and a Triplet loss function and (2) the feasibility of end-to-end learning which includes the feature embedding network and attention weight computation. The feature embedding network has a positive effect on the attention weight computation by using combined loss function and end-to-end learning. To demonstrate the effectiveness of our proposed method, extensive and comparative experiments have been carried out to evaluate our method on IJB-A dataset with their standard evaluation protocols. Our proposed method represented better or comparable recognition rate compared to other state-of-the-art video FR methods.

• Smart Media Journal
• /
• v.12 no.3
• /
• pp.30-37
• /
• 2023

Alzheimer's disease (AD) symptoms are being treated by early diagnosis, where we can only slow the symptoms and research is still undergoing. In consideration, using T1-weighted images several classification models are proposed in Machine learning to identify AD. In this paper, we consider the improvised feature selection, to reduce the complexity by using wrapping techniques and Restricted Boltzmann Machine (RBM). This present work used the subcortical and cortical features of 278 subjects from the ADNI dataset to identify AD and sMRI. Multi-class classification is used for the experiment i.e., AD, EMCI, LMCI, HC. The proposed feature selection consists of Forward feature selection, Backward feature selection, and Combined PCA & RBM. Forward and backward feature selection methods use an iterative method starting being no features in the forward feature selection and backward feature selection with all features included in the technique. PCA is used to reduce the dimensions and RBM is used to select the best feature without interpreting the features. We have compared the three models with PCA to analysis. The following experiment shows that combined PCA &RBM, and backward feature selection give the best accuracy with respective classification model RF i.e., 88.65, 88.56% respectively.

• Korean Journal of Remote Sensing
• /
• v.31 no.3
• /
• pp.205-214
• /
• 2015

In this study, a novel land-cover classification framework for multi-temporal SAR data is presented that can combine multiple features extracted through data transforms and multiple classifiers. At first, data transforms using principle component analysis (PCA) and 3D wavelet transform are applied to multi-temporal SAR dataset for extracting new features which were different from original dataset. Then, three different classifiers including maximum likelihood classifier (MLC), neural network (NN) and support vector machine (SVM) are applied to three different dataset including data transform based features and original backscattering coefficients, and as a result, the diverse preliminary classification results are generated. These results are combined via a majority voting rule to generate a final classification result. From an experiment with a multi-temporal ENVISAT ASAR dataset, every preliminary classification result showed very different classification accuracy according to the used feature and classifier. The final classification result combining nine preliminary classification results showed the best classification accuracy because each preliminary classification result provided complementary information on land-covers. The improvement of classification accuracy in this study was mainly attributed to the diversity from combining not only different features based on data transforms, but also different classifiers. Therefore, the land-cover classification framework presented in this study would be effectively applied to the classification of multi-temporal SAR data and also be extended to multi-sensor remote sensing data fusion.

• Journal of Internet Computing and Services
• /
• v.23 no.4
• /
• pp.45-55
• /
• 2022

The size of the cryptocurrency market is growing. For example, market capitalization of bitcoin exceeded 500 trillion won. Accordingly, many studies have been conducted to predict the price of cryptocurrency, and most of them have similar methodology of predicting stock prices. However, unlike stock price predictions, machine learning become best model in cryptocurrency price predictions, conceptually cryptocurrency has no passive income from ownership, and statistically, cryptocurrency has at least three times higher liquidity than stocks. Thats why we argue that a methodology different from stock price prediction should be applied to cryptocurrency price prediction studies. We propose Reverse Walk-forward Validation (RWFV), which modifies Walk-forward Validation (WFV). Unlike WFV, RWFV measures accuracy for Validation by pinning the Validation dataset directly in front of the Test dataset in time series, and gradually increasing the size of the Training dataset in front of it in time series. Train data were cut according to the size of the Train dataset with the highest accuracy among all measured Validation accuracy, and then combined with Validation data to measure the accuracy of the Test data. Logistic regression analysis and Support Vector Machine (SVM) were used as the analysis model, and various algorithms and parameters such as L1, L2, rbf, and poly were applied for the reliability of our proposed RWFV. As a result, it was confirmed that all analysis models showed improved accuracy compared to existing studies, and on average, the accuracy increased by 1.23%p. This is a significant improvement in accuracy, given that most of the accuracy of cryptocurrency price prediction remains between 50% and 60% through previous studies.

• Journal of Intelligence and Information Systems
• /
• v.22 no.1
• /
• pp.139-157
• /
• 2016

Bankruptcy involves considerable costs, so it can have significant effects on a country's economy. Thus, bankruptcy prediction is an important issue. Over the past several decades, many researchers have addressed topics associated with bankruptcy prediction. Early research on bankruptcy prediction employed conventional statistical methods such as univariate analysis, discriminant analysis, multiple regression, and logistic regression. Later on, many studies began utilizing artificial intelligence techniques such as inductive learning, neural networks, and case-based reasoning. Currently, ensemble models are being utilized to enhance the accuracy of bankruptcy prediction. Ensemble classification involves combining multiple classifiers to obtain more accurate predictions than those obtained using individual models. Ensemble learning techniques are known to be very useful for improving the generalization ability of the classifier. Base classifiers in the ensemble must be as accurate and diverse as possible in order to enhance the generalization ability of an ensemble model. Commonly used methods for constructing ensemble classifiers include bagging, boosting, and random subspace. The random subspace method selects a random feature subset for each classifier from the original feature space to diversify the base classifiers of an ensemble. Each ensemble member is trained by a randomly chosen feature subspace from the original feature set, and predictions from each ensemble member are combined by an aggregation method. The k-nearest neighbors (KNN) classifier is robust with respect to variations in the dataset but is very sensitive to changes in the feature space. For this reason, KNN is a good classifier for the random subspace method. The KNN random subspace ensemble model has been shown to be very effective for improving an individual KNN model. The k parameter of KNN base classifiers and selected feature subsets for base classifiers play an important role in determining the performance of the KNN ensemble model. However, few studies have focused on optimizing the k parameter and feature subsets of base classifiers in the ensemble. This study proposed a new ensemble method that improves upon the performance KNN ensemble model by optimizing both k parameters and feature subsets of base classifiers. A genetic algorithm was used to optimize the KNN ensemble model and improve the prediction accuracy of the ensemble model. The proposed model was applied to a bankruptcy prediction problem by using a real dataset from Korean companies. The research data included 1800 externally non-audited firms that filed for bankruptcy (900 cases) or non-bankruptcy (900 cases). Initially, the dataset consisted of 134 financial ratios. Prior to the experiments, 75 financial ratios were selected based on an independent sample t-test of each financial ratio as an input variable and bankruptcy or non-bankruptcy as an output variable. Of these, 24 financial ratios were selected by using a logistic regression backward feature selection method. The complete dataset was separated into two parts: training and validation. The training dataset was further divided into two portions: one for the training model and the other to avoid overfitting. The prediction accuracy against this dataset was used to determine the fitness value in order to avoid overfitting. The validation dataset was used to evaluate the effectiveness of the final model. A 10-fold cross-validation was implemented to compare the performances of the proposed model and other models. To evaluate the effectiveness of the proposed model, the classification accuracy of the proposed model was compared with that of other models. The Q-statistic values and average classification accuracies of base classifiers were investigated. The experimental results showed that the proposed model outperformed other models, such as the single model and random subspace ensemble model.

• Proceedings of the Korea Institutes of Information Security and Cryptology Conference
• /
• 2006.06a
• /
• pp.649-653
• /
• 2006

In this paper, we propose a hybrid feature extraction method in which Principal Components Analysis is combined with optimized k-Means clustering technique. Our approach hierarchically reduces the redundancy of features with high explanation in principal components analysis for choosing a good subset of features critical to improve the performance of classifiers. Based on this result, we evaluate the performance of intrusion detection by using Support Vector Machine and a nonparametric approach based on k-Nearest Neighbor over data sets with reduced features. The Experiment results with KDD Cup 1999 dataset show several advantages in terms of computational complexity and our method achieves significant detection rate which shows possibility of detecting successfully attacks.

• ETRI Journal
• /
• v.38 no.3
• /
• pp.494-501
• /
• 2016

Sentiment analysis incorporates natural language processing and artificial intelligence and has evolved as an important research area. Sentiment analysis on product reviews has been used in widespread applications to improve customer retention and business processes. In this paper, we propose a method for performing an intensified sentiment analysis on customer product reviews. The method involves the extraction of two feature sets from each of the given customer product reviews, a set of acoustic features (representing emotions) and a set of lexical features (representing sentiments). These sets are then combined and used in a supervised classifier to predict the sentiments of customers. We use an audio speech dataset prepared from Amazon product reviews and downloaded from the YouTube portal for the purposes of our experimental evaluations.

• ETRI Journal
• /
• v.42 no.3
• /
• pp.411-419
• /
• 2020

In real-world intelligent transportation systems, accuracy in vehicle license plate detection and recognition is considered quite critical. Many algorithms have been proposed for still images, but their accuracy on actual videos is not satisfactory. This stems from several problematic conditions in videos, such as vehicle motion blur, variety in viewpoints, outliers, and the lack of publicly available video datasets. In this study, we focus on these challenges and propose a license plate detection and recognition scheme for videos based on a temporal matching prior network. Specifically, to improve the robustness of detection and recognition accuracy in the presence of motion blur and outliers, forward and bidirectional matching priors between consecutive frames are properly combined with layer structures specifically designed for plate detection. We also built our own video dataset for the deep training of the proposed network. During network training, we perform data augmentation based on image rotation to increase robustness regarding the various viewpoints in videos.

• ETRI Journal
• /
• v.37 no.3
• /
• pp.502-511
• /
• 2015

Intrusion detection plays a key role in detecting attacks over networks, and due to the increasing usage of Internet services, several security threats arise. Though an intrusion detection system (IDS) detects attacks efficiently, it also generates a large number of false alerts, which makes it difficult for a system administrator to identify attacks. This paper proposes automatic fuzzy rule generation combined with a Wiener filter to identify attacks. Further, to optimize the results, simplified swarm optimization is used. After training a large dataset, various fuzzy rules are generated automatically for testing, and a Wiener filter is used to filter out attacks that act as noisy data, which improves the accuracy of the detection. By combining automatic fuzzy rule generation with a Wiener filter, an IDS can handle intrusion detection more efficiently. Experimental results, which are based on collected live network data, are discussed and show that the proposed method provides a competitively high detection rate and a reduced false alarm rate in comparison with other existing machine learning techniques.

• Mycobiology
• /
• v.48 no.4
• /
• pp.321-325
• /
• 2020

A Colletotrichum species was isolated from leaves of Cymbidium exhibiting symptoms of anthracnose. In this study, the isolates obtained were identified based on recent taxonomic approaches for the genus Colletotrichum. The identity of the causal pathogen was confirmed using morphological data and phylogenetic analysis of combined multi-gene dataset (internal transcribed spacer, glyceraldehyde 3-phosphate dehydrogenase, chitin synthase-1, actin, histone3, beta-tubulin, and calmodulin). Pathogenicity testing revealed that the isolates were pathogenic to Cymbidium. Based on these results, the fungal pathogen occurring on Cymbidium orchids was identified as Colletotrichum cymbidiicola, which is a newly recorded species in Korea.