This page has only limited features, please log in for full access.
Sung Bum Pan received the B.S., M.S., and Ph.D. degrees in electronics engineering from Sogang University, South Korea, in 1991, 1995, and 1999, respectively. He was a Team Leader with the Biometric Technology Research Team, ETRI, from 1999 to 2005. He is currently a Professor with Chosun University. His current research interests include biometrics, security, and VLSI architectures for real-time image processing.
Electromyogram (EMG) signals cannot be forged and have the advantage of being able to change the registered data as they are characterized by the waveform, which varies depending on the gesture. In this paper, a two-step biometrics method was proposed using EMG signals based on a convolutional neural network–long short-term memory (CNN-LSTM) network. After preprocessing of the EMG signals, the time domain features and LSTM network were used to examine whether the gesture matched, and single biometrics was performed if the gesture matched. In single biometrics, EMG signals were converted into a two-dimensional spectrogram, and training and classification were performed through the CNN-LSTM network. Data fusion of the gesture recognition and single biometrics was performed in the form of an AND. The experiment used Ninapro EMG signal data as the proposed two-step biometrics method, and the results showed 83.91% gesture recognition performance and 99.17% single biometrics performance. In addition, the false acceptance rate (FAR) was observed to have been reduced by 64.7% through data fusion.
Jin-Su Kim; Min-Gu Kim; Sung-Bum Pan. Two-Step Biometrics Using Electromyogram Signal Based on Convolutional Neural Network-Long Short-Term Memory Networks. Applied Sciences 2021, 11, 6824 .
AMA StyleJin-Su Kim, Min-Gu Kim, Sung-Bum Pan. Two-Step Biometrics Using Electromyogram Signal Based on Convolutional Neural Network-Long Short-Term Memory Networks. Applied Sciences. 2021; 11 (15):6824.
Chicago/Turabian StyleJin-Su Kim; Min-Gu Kim; Sung-Bum Pan. 2021. "Two-Step Biometrics Using Electromyogram Signal Based on Convolutional Neural Network-Long Short-Term Memory Networks." Applied Sciences 11, no. 15: 6824.
Electrocardiogram (ECG) signals are time series data that are acquired by time change. A problem with these signals is that comparison data that have the same size as the registration data must be acquired every time. A network model of an auxiliary classifier based generative adversarial neural network that is capable of generating synthetic ECG signals is proposed to resolve the data size inconsistency problem. After constructing comparison data with various combinations of the real and generated synthetic ECG signal cycles, a user recognition experiment was performed by applying them to an ensemble network of parallel structure. Recognition performance of 98.5% was demonstrated when five cycles of real ECG signals were used. Moreover, 98.7% and 97% accuracies were provided when the first cycle of synthetic ECG signals and the fourth cycle of real ECG signals were repetitively used as the last cycle, respectively, in addition to the four cycles of real ECG. When two cycles of synthetic ECG signals were used with three cycles of real ECG signals, 97.2% accuracy was shown. When the last third cycle was repeatedly used with the three cycles of real ECG signals, the accuracy was 96%, which was 1.2% lower than the performance obtained while using the synthetic ECG. Therefore, even if the size of the registration data and that of the comparison data are not consistent, the generated synthetic ECG signals can be applied to a real life environment, because a high recognition performance is demonstrated when they are applied to an ensemble network of parallel structure.
Min-Gu Kim; Sung Pan. A Study on User Recognition Using the Generated Synthetic Electrocardiogram Signal. Sensors 2021, 21, 1887 .
AMA StyleMin-Gu Kim, Sung Pan. A Study on User Recognition Using the Generated Synthetic Electrocardiogram Signal. Sensors. 2021; 21 (5):1887.
Chicago/Turabian StyleMin-Gu Kim; Sung Pan. 2021. "A Study on User Recognition Using the Generated Synthetic Electrocardiogram Signal." Sensors 21, no. 5: 1887.
최근 생체정보를 이용한 사용자 인식은 위 · 변조 사건, 사고로 사회적 이슈가 되어 생체신호를 이용한 사용자 인식 연구가 활발히 진행 중이다. 생체신호는 신체 내부에서 발생하는 전기적 신호로 행동학적 특징에 따라 개인의 고유한 신호를 발생한다. 생체신호는 대표적으로 근전도, 심전도, 뇌전도 등이 있다. 생체신호 중 근전도 신호는 개인의 고유한 근력 세기에 따라 서로 다른 신호 패턴으로 측정되는 특징을 이용하여 사용자 인식 분야에서 적용되고 있다. 본 논문에서는 근전도 신호를 이용한 연구 및 응용 분야들을 분석하고 근전도 기반 사용자 인식 시스템을 분석하였다. 그리고 전처리된 1차원 근전도 신호를 2차원 근전도 스펙트로그램 이미지로 변환하고 CNN을 이용하여 특징 추출 후 사용자를 최종 인식하여 사용자 인식 시스템의 성능을 분석한다. 실험결과 40명에 대한 사용자 인식은 12채널을 이용하고 STFT의 Window length R이 256일 때 95.4%(±1.7%)로 분석되었다.
Jae Myung Kim; Gyu Ho Choi; Jin Su Kim; Sung Bum Pan. User Recognition using Electromyogram 2D Spectrogram Images based on CNN. The Journal of Korean Institute of Information Technology 2021, 19, 107 -117.
AMA StyleJae Myung Kim, Gyu Ho Choi, Jin Su Kim, Sung Bum Pan. User Recognition using Electromyogram 2D Spectrogram Images based on CNN. The Journal of Korean Institute of Information Technology. 2021; 19 (1):107-117.
Chicago/Turabian StyleJae Myung Kim; Gyu Ho Choi; Jin Su Kim; Sung Bum Pan. 2021. "User Recognition using Electromyogram 2D Spectrogram Images based on CNN." The Journal of Korean Institute of Information Technology 19, no. 1: 107-117.
In this report, the study of non-fiducial based approaches for Electrocardiogram(ECG) biometric authentication is examined, and several excessive techniques are proposed to perform comparative experiments for evaluating the best possible approach for all the classification tasks. Non-fiducial methods are designed to extract the discriminative information of a signal without annotating fiducial points. However, this process requires peak detection to identify a heartbeat signal. Based on recent studies that usually rely on heartbeat segmentation, QRS detection is required, and the process can be complicated for ECG signals for which the QRS complex is absent. Thus, many studies only conduct biometric authentication tasks on ECG signals with QRS complexes, and are hindered by similar limitations. To overcome this issue, we proposed a data-independent acquisition method to facilitate highly generalizable signal processing and feature learning processes. This is achieved by enhancing random segmentation to avoid complicated fiducial feature extraction, along with auto-correlation to eliminate the phase difference due to random segmentation. Subsequently, a bidirectional recurrent neural network (RNN) with long short-term memory (LSTM) deep networks is utilized to automatically learn the features associated with the signal and to perform an authentication task. The experimental results suggest that the proposed data-independent approach using a BLSTM network achieves a relatively high classification accuracy for every dataset relative to the compared techniques. Moreover, it exhibited a significantly higher accuracy rate in experiments using ECG signals without the QRS complex. The results also revealed that data-dependent methods can only perform well for specified data types and amendments of data variations, whereas the presented approach can also be considered for generalization to other quasi-periodical biometric signal-based classification tasks in future studies.
Htet Myet Lynn; Pankoo Kim; Sung Bum Pan. Data Independent Acquisition Based Bi-Directional Deep Networks for Biometric ECG Authentication. Applied Sciences 2021, 11, 1125 .
AMA StyleHtet Myet Lynn, Pankoo Kim, Sung Bum Pan. Data Independent Acquisition Based Bi-Directional Deep Networks for Biometric ECG Authentication. Applied Sciences. 2021; 11 (3):1125.
Chicago/Turabian StyleHtet Myet Lynn; Pankoo Kim; Sung Bum Pan. 2021. "Data Independent Acquisition Based Bi-Directional Deep Networks for Biometric ECG Authentication." Applied Sciences 11, no. 3: 1125.
Driver-centered infotainment and telematics services are provided for intelligent vehicles that improve driver convenience. Driver-centered services are performed after identification, and a biometrics system using bio-signals is applied. The electrocardiogram (ECG) signal acquired in the driving environment needs to be normalized because the intensity of noise is strong because the driver’s motion artifact is included. Existing time, frequency, and phase normalization methods have a problem of distorting P, QRS Complexes, and T waves, which are morphological features of an ECG, or normalizing to signals containing noise. In this paper, we propose an adaptive threshold filter-based driver identification system to solve the problem of distortion of the ECG morphological features when normalized and the motion artifact noise of the ECG that causes the identification performance deterioration in the driving environment. The experimental results show that the proposed method improved the average similarity compared to the results without normalization. The identification performance was also improved compared to the results before normalization.
Gyu Ho Choi; Kiho Lim; Sung Bum Pan. Driver Identification System Using Normalized Electrocardiogram Based on Adaptive Threshold Filter for Intelligent Vehicles. Sensors 2020, 21, 202 .
AMA StyleGyu Ho Choi, Kiho Lim, Sung Bum Pan. Driver Identification System Using Normalized Electrocardiogram Based on Adaptive Threshold Filter for Intelligent Vehicles. Sensors. 2020; 21 (1):202.
Chicago/Turabian StyleGyu Ho Choi; Kiho Lim; Sung Bum Pan. 2020. "Driver Identification System Using Normalized Electrocardiogram Based on Adaptive Threshold Filter for Intelligent Vehicles." Sensors 21, no. 1: 202.
Although biometrics systems using an electrocardiogram (ECG) have been actively researched, there is a characteristic that the morphological features of the ECG signal are measured differently depending on the measurement environment. In general, post-exercise ECG is not matched with the morphological features of the pre-exercise ECG because of the temporary tachycardia. This can degrade the user recognition performance. Although normalization studies have been conducted to match the post- and pre-exercise ECG, limitations related to the distortion of the P wave, QRS complexes, and T wave, which are morphological features, often arise. In this paper, we propose a method for matching pre- and post-exercise ECG cycles based on time and frequency fusion normalization in consideration of morphological features and classifying users with high performance by an optimized system. One cycle of post-exercise ECG is expanded by linear interpolation and filtered with an optimized frequency through the fusion normalization method. The fusion normalization method aims to match one post-exercise ECG cycle to one pre-exercise ECG cycle. The experimental results show that the average similarity between the pre- and post-exercise states improves by 25.6% after normalization, for 30 ECG cycles. Additionally, the normalization algorithm improves the maximum user recognition performance from 96.4 to 98%.
Gyu Ho Choi; Hoon Ko; Witold Pedrycz; Amit Kumar Singh; Sung Bum Pan. Recognition System Using Fusion Normalization Based on Morphological Features of Post-Exercise ECG for Intelligent Biometrics. Sensors 2020, 20, 7130 .
AMA StyleGyu Ho Choi, Hoon Ko, Witold Pedrycz, Amit Kumar Singh, Sung Bum Pan. Recognition System Using Fusion Normalization Based on Morphological Features of Post-Exercise ECG for Intelligent Biometrics. Sensors. 2020; 20 (24):7130.
Chicago/Turabian StyleGyu Ho Choi; Hoon Ko; Witold Pedrycz; Amit Kumar Singh; Sung Bum Pan. 2020. "Recognition System Using Fusion Normalization Based on Morphological Features of Post-Exercise ECG for Intelligent Biometrics." Sensors 20, no. 24: 7130.
In this paper, the method to overcome the limitations of the existing three-dimensional golf swing analysis system by using deep learning technology, and analyze the three-dimensional quantitative information through sequence images acquired with a single camera is studied. In this paper, CNN was used to extract the appropriate features from the image of the golf frontal swing sequence, and a regression model based on Bi-LSTM was used to predict the correct information in each sequence. This classifies the major swing section, and analyzes the quantitative status of the twisting angles of the upper body, head, shoulder and pelvis for body-sway, head-up and X-factor analysis. For the experiment, in this paper, a total of 520 times swing data were obtained using no. 1 wood club and no. 7 iron club from five subjects. In the major swing section classification experiment, each swing section was classified with an average accuracy of about 95.44%. Quantitative analysis results from each analysis model showed that the upper body motion prediction RMSE averaged 4.23 degrees, the head motion prediction RMSE averaged 5.18 degrees, and the shoulder and pelvis twisting angle prediction RMSE averaged 3.86 degrees. As a result, it was confirmed that a three-dimensional quantitative analysis based on sequence images is possible.
Kyeong-Ri Ko; Sung Bum Pan. CNN and bi-LSTM based 3D golf swing analysis by frontal swing sequence images. Multimedia Tools and Applications 2020, 80, 8957 -8972.
AMA StyleKyeong-Ri Ko, Sung Bum Pan. CNN and bi-LSTM based 3D golf swing analysis by frontal swing sequence images. Multimedia Tools and Applications. 2020; 80 (6):8957-8972.
Chicago/Turabian StyleKyeong-Ri Ko; Sung Bum Pan. 2020. "CNN and bi-LSTM based 3D golf swing analysis by frontal swing sequence images." Multimedia Tools and Applications 80, no. 6: 8957-8972.
As the importance of providing personalized services increases, various studies on personalized recommendation systems are actively being conducted. Among the many methods used for recommendation systems, the most widely used is collaborative filtering. However, this method has lower accuracy because recommendations are limited to using quantitative information, such as user ratings or amount of use. To address this issue, many studies have been conducted to improve the accuracy of the recommendation system by using other types of information, in addition to quantitative information. Although conducting sentiment analysis using reviews is popular, previous studies show the limitation that results of sentiment analysis cannot be directly reflected in recommendation systems. Therefore, this study aims to quantify the sentiments presented in the reviews and reflect the results to the ratings; that is, this study proposes a new algorithm that quantifies the sentiments of user-written reviews and converts them into quantitative information, which can be directly reflected in recommendation systems. To achieve this, the user reviews, which are qualitative information, must first be quantified. Thus, in this study, sentiment scores are calculated through sentiment analysis by using a text mining technique. The data used herein are from movie reviews. A domain-specific sentiment dictionary was constructed, and then based on the dictionary, sentiment scores of the reviews were calculated. The collaborative filtering of this study, which reflected the sentiment scores of user reviews, was verified to demonstrate its higher accuracy than the collaborative filtering using the traditional method, which reflects only user rating data. To overcome the limitations of the previous studies that examined the sentiments of users based only on user rating data, the method proposed in this study successfully enhanced the accuracy of the recommendation system by precisely reflecting user opinions through quantified user reviews. Based on the findings of this study, the recommendation system accuracy is expected to improve further if additional analysis can be performed.
Tae-Yeun Kim; Sung Bum Pan; Sung-Hwan Kim. Sentiment Digitization Modeling for Recommendation System. Sustainability 2020, 12, 5191 .
AMA StyleTae-Yeun Kim, Sung Bum Pan, Sung-Hwan Kim. Sentiment Digitization Modeling for Recommendation System. Sustainability. 2020; 12 (12):5191.
Chicago/Turabian StyleTae-Yeun Kim; Sung Bum Pan; Sung-Hwan Kim. 2020. "Sentiment Digitization Modeling for Recommendation System." Sustainability 12, no. 12: 5191.
ECG data are biosignals with unique characteristics that can be obtained regardless of time and space constraints. Owing to these advantages, they have been widely used for not only diagnosing diseases but also recognizing people. Numerous studies have been conducted and various feature vectors from a large amount of data have been suggested to improve recognition performance. The key to extracting feature vectors is to extract differences in one-dimensional ECG signals without loss in order to recognize human identity. In this paper, we propose new feature vectors based on fiducial points. These feature vectors have simple and clear shapes that combine temporal and amplitude information. The discriminator operating in the proposed human identification system measures distance-based similarity. This method alleviates computational burden and enables the human identification system to run in real time. Based on the system, we conducted a number of recognition experiments. The experimental results proved that the proposed feature vectors are valid information that represents significant differences between individuals. In the experiments with 100 subjects, we obtained a recognition rate of over 94% when two or more than two heartbeat signals were used, and confirmed that as the number of input heartbeats increased the performance also improved proportionally.
Eunsang Bak; Gyu-Ho Choi; Sung Bum Pan. ECG-Based Human Identification System by Temporal-Amplitude Combined Feature Vectors. IEEE Access 2020, 8, 42217 -42230.
AMA StyleEunsang Bak, Gyu-Ho Choi, Sung Bum Pan. ECG-Based Human Identification System by Temporal-Amplitude Combined Feature Vectors. IEEE Access. 2020; 8 (99):42217-42230.
Chicago/Turabian StyleEunsang Bak; Gyu-Ho Choi; Sung Bum Pan. 2020. "ECG-Based Human Identification System by Temporal-Amplitude Combined Feature Vectors." IEEE Access 8, no. 99: 42217-42230.
Research on electrocardiogram (ECG) signals has been actively undertaken to assess their value as a next generation user recognition technology, because they require no stimulation and are robust against forgery and modification. However, even within the same user, the heart rate and waveform of ECG signals will vary depending on physical activity, mental effects, and measurement time. Therefore, when data acquired across changes in the user state is used as registered data, an overfitting problem occurs due to data generalization, which degrades the recognition performance for newly acquired data. Therefore, in this paper, we propose parallel ensemble networks to solve the overfitting problem and prevent the degradation. First, ECG signals acquired in various environments are used as the input data of parallel convolutional neural networks (CNNs). Each CNN is set up with different parameters to detect different features. The ECG signals outputted from each network are classified for each subject, and then fused into one database to be used as registered data for re-training. Instead of fusing all the output signals from each network, only the ECG signals of Top-3 networks showing excellent performance are fused and composed of registered data. The reconstructed registered data are used for user recognition, by re-training with time independent comparison data in the CNN. The experimental results of comparing the proposed parallel ensemble networks with those of previous studies using the self-acquired actual ECG signals show that the proposed method achieves recognition performance higher than the previous studies, with an accuracy rate of 98.5%.
Min-Gu Kim; Chang Choi; Sung Bum Pan. Ensemble Networks for User Recognition in Various Situations Based on Electrocardiogram. IEEE Access 2020, 8, 36527 -36535.
AMA StyleMin-Gu Kim, Chang Choi, Sung Bum Pan. Ensemble Networks for User Recognition in Various Situations Based on Electrocardiogram. IEEE Access. 2020; 8 (99):36527-36535.
Chicago/Turabian StyleMin-Gu Kim; Chang Choi; Sung Bum Pan. 2020. "Ensemble Networks for User Recognition in Various Situations Based on Electrocardiogram." IEEE Access 8, no. 99: 36527-36535.
기존 개인 식별 방법은 다양한 범죄에 취약한 문제점이 있어 이를 보완하기 위해 신체 내부 특징인 바이오 신호를 활용한 연구가 진행되고 있다. 그중 심전도 신호는 심장의 크기와 위치에 따라 개인마다 고유한 특성이 있어 개인 식별에 적합하며, 딥 러닝과 접목하여 많은 연구가 진행되고 있다. 본 논문에서는 2차원 심전도 이미지를 이용한 사전 학습된 네트워크 모델에 따른 개인 식별 성능을 분석한다. 심전도 데이터 세트를 학습시키기 위한 사전 학습된 네트워크 모델은 Inception, ResNet의 11개 네트워크를 사용한다. 네트워크의 학습데이터는 심전도 신호의 한 주기로 생성한 2차원 이미지 데이터를 이용하며, 실험은 학습 횟수를 변경하며 진행한다. 사전 학습된 네트워크 모델을 이용한 심전도 신호 기반 개인 식별 실험 결과 Inception 네트워크에 비해 ResNet 네트워크의 성능이 높게 나타났으며, Inception 네트워크에선 Inception-ResNet-V2가 96.18%, ResNet 네트워크에선 ResNet-V2-152가 99.12%의 성능으로 가장 높음을 확인하였다.
Jin Su Kim; Sung Huck Kim; Sung Bum Pan. Electrocardiogram Signal Based Personal Identification Performance Analysis Using Pre-trained Network Model. The Journal of Korean Institute of Information Technology 2020, 18, 107 -114.
AMA StyleJin Su Kim, Sung Huck Kim, Sung Bum Pan. Electrocardiogram Signal Based Personal Identification Performance Analysis Using Pre-trained Network Model. The Journal of Korean Institute of Information Technology. 2020; 18 (1):107-114.
Chicago/Turabian StyleJin Su Kim; Sung Huck Kim; Sung Bum Pan. 2020. "Electrocardiogram Signal Based Personal Identification Performance Analysis Using Pre-trained Network Model." The Journal of Korean Institute of Information Technology 18, no. 1: 107-114.
Taking care of individual pigs is important in the management of a group-housed pig farm. However, this is nearly impossible in a large-scale pig farm owing to the shortage of farm workers. Therefore, we propose an automatic monitoring method in this study to solve the management problem of a large-scale pig farm. Particularly, we aim to detect undergrown pigs in group-housed pig rooms by using deep-learning-based computer vision techniques. Because the typical deep learning techniques require a large computational overhead (i.e., Mask-R-CNN), fast and accurate detection of undergrown pigs on an IoT-based embedded device is very challenging. We first obtain the video monitoring data of grouphoused pigs by using a top-view camera that is installed in the pig room, and then detect each moving pig by combining image processing and deep learning techniques. Gaussian Mixture Model is used to detect moving frames and moving objects. In embedded device implementations, by applying deep learning (i.e., TinyYOLO3) only to a few frames with a large number of pixel changes, embedded GPUs can be used efficiently, satisfying the real-time requirement. As a subsequent step, we check the acceptable conditions of the posture and separability from each video frame of the continuous video stream. Finally, to compute the relative size of each pig quickly and accurately, we develop image processing steps to complement the result of deep learning with minimum computational overhead. Furthermore, by pipelining the CPU and GPU steps of a continuous video stream, we can hide the additional image processing time. Based on the experimental results obtained from an embedded device, we confirm that the proposed method can automatically detect undergrown pigs in real-time, by working as an early warning system without any human inspection or measurement of actual weight by a farm worker.
Sungju Lee; HansE Ahn; Jihyun Seo; Yongwha Chung; Daihee Park; Sungbum Pan. Practical Monitoring of Undergrown Pigs for IoT-Based Large-Scale Smart Farm. IEEE Access 2019, 7, 173796 -173810.
AMA StyleSungju Lee, HansE Ahn, Jihyun Seo, Yongwha Chung, Daihee Park, Sungbum Pan. Practical Monitoring of Undergrown Pigs for IoT-Based Large-Scale Smart Farm. IEEE Access. 2019; 7 (99):173796-173810.
Chicago/Turabian StyleSungju Lee; HansE Ahn; Jihyun Seo; Yongwha Chung; Daihee Park; Sungbum Pan. 2019. "Practical Monitoring of Undergrown Pigs for IoT-Based Large-Scale Smart Farm." IEEE Access 7, no. 99: 173796-173810.
There are many blogs that recommend places and foods and on the web. In addition, there are various fake news that provide false information. They both are written by a blogger; bloggers can write on any topic of their choice. Web visitors read these blogs and decide if place or food item is satisfactory. This implies that the decision is based on the blogger’s prejudice. This is not objective because all the decisions depend on the blogger’s disposition. Other visitors, who had followed the bloggerâĂŹs recommendation, may have disagree with the blogger. To avoid this conflict, all the words and sentences in the posts must be analyzed objectively. All entries such as direction, address, excessive compliments, and monophonic are analyzed. This study also analyzed the entries to see their correlation; finally, it can make the decision if a blog is trustable with an anomaly sign.
Hoon Ko; Libor Mesicek; Jong Youl Hong; Soon Sim Yeo; Sung Bum Pan; Pankoo Kim. Blog Reliability Analysis With Conflicting Interests of Contexts in the Extended Branch for Cyber-Security. IEEE Access 2019, 7, 143693 -143698.
AMA StyleHoon Ko, Libor Mesicek, Jong Youl Hong, Soon Sim Yeo, Sung Bum Pan, Pankoo Kim. Blog Reliability Analysis With Conflicting Interests of Contexts in the Extended Branch for Cyber-Security. IEEE Access. 2019; 7 (99):143693-143698.
Chicago/Turabian StyleHoon Ko; Libor Mesicek; Jong Youl Hong; Soon Sim Yeo; Sung Bum Pan; Pankoo Kim. 2019. "Blog Reliability Analysis With Conflicting Interests of Contexts in the Extended Branch for Cyber-Security." IEEE Access 7, no. 99: 143693-143698.
In this paper, we propose a deep Recurrent Neural Networks (RNNs) based on Gated Recurrent Unit (GRU) in a bidirectional manner (BGRU) for human identification from electrocardiogram (ECG) based biometrics, a classification task which aims to identify a subject from a given time-series sequential data. Despite having a major issue in traditional RNN networks which they learn representations from previous time sequences, bidirectional is designed to learn the representations from future time steps which enables for better understanding of context, and eliminate ambiguity. Moreover, GRU cell in RNNs deploys an update gate and a reset gate in a hidden state layer which is computationally efficient than a usual LSTM network due to the reduction of gates. The experimental results suggest that our proposed BGRU model, the combination of RNN with GRU cell unit in bidirectional manner, achieved a high classification accuracy of 98.55%. Various neural network architectures with different parameters are also evaluated for different approaches, including one-dimensional Convolutional Neural Network (1D-CNN), and traditional RNNs with LSTM and GRU for non-fiducial approach. The proposed models were evaluated with two publicly available datasets: ECG-ID Database (ECGID) and MIT-BIH Arrhythmia Database (MITDB). This paper is expected to demonstrate the feasibility and effectiveness of applying various deep learning approaches to biometric identification and also evaluate the effect of network performance on classification accuracy according to the changes in percentage of training dataset.
Htet Myet Lynn; Sung Bum Pan; Pankoo Kim. A Deep Bidirectional GRU Network Model for Biometric Electrocardiogram Classification Based on Recurrent Neural Networks. IEEE Access 2019, 7, 145395 -145405.
AMA StyleHtet Myet Lynn, Sung Bum Pan, Pankoo Kim. A Deep Bidirectional GRU Network Model for Biometric Electrocardiogram Classification Based on Recurrent Neural Networks. IEEE Access. 2019; 7 (99):145395-145405.
Chicago/Turabian StyleHtet Myet Lynn; Sung Bum Pan; Pankoo Kim. 2019. "A Deep Bidirectional GRU Network Model for Biometric Electrocardiogram Classification Based on Recurrent Neural Networks." IEEE Access 7, no. 99: 145395-145405.
Nowadays, with rapid advancement of both the upcoming 5G architecture construction and emerging Internet of Things (IoT) scenarios, Device-to-Device (D2D) communication provides a novel paradigm for mobile networking. By facilitating continuous and high data rate services between physically proximate devices without interconnection with access points (AP) or service network (SN), spectral efficiency of the 5G network can be drastically increased. However, due to its inherent open wireless communicating features, security issues and privacy risks in D2D communication remain unsolved in spite of its benefits and prosperous future. Hence, proper D2D authentication mechanisms among the D2D entities are of great significance. Moreover, the increasing proliferation of smartphones enables seamlessly biometric sensor data collecting and processing, which highly correspond to the user’s unique behavioral characteristics. For the above consideration, we present a secure certificateless D2D authenticating mechanism intended for extreme scenarios in this paper. In the assumption, the key updating mechanism only requires a small modification in the SN side, while the decryption information of user equipment (UEs) remains constant as soon as the UEs are validated. Note that a symmetric key mechanism is adopted for the further data transmission. Additionally, the user activities data from smartphone sensors are analyzed for continuous authentication, which is periodically conducted after the initial validation. Note that in the assumed scenario, most of the UEs are out of the effective range of cellular networks. In this case, the UEs are capable of conducting data exchange without cellular connection. Security analysis demonstrates that the proposed scheme can provide adequate security properties as well as resistance to various attacks. Furthermore, performance analysis proves that the proposed scheme is efficient compared with state-of-the-art D2D authentication schemes.
Haowen Tan; Yuanzhao Song; Shichang Xuan; Sungbum Pan; Ilyong Chung. Secure D2D Group Authentication Employing Smartphone Sensor Behavior Analysis. Symmetry 2019, 11, 969 .
AMA StyleHaowen Tan, Yuanzhao Song, Shichang Xuan, Sungbum Pan, Ilyong Chung. Secure D2D Group Authentication Employing Smartphone Sensor Behavior Analysis. Symmetry. 2019; 11 (8):969.
Chicago/Turabian StyleHaowen Tan; Yuanzhao Song; Shichang Xuan; Sungbum Pan; Ilyong Chung. 2019. "Secure D2D Group Authentication Employing Smartphone Sensor Behavior Analysis." Symmetry 11, no. 8: 969.
Personal identification method using the Electrocardiogram (ECG) signal is an active research area since the ECG signal cannot be forged and can be acquired without active awareness by the subject. In this paper, we propose a personal recognition system using the 2-D coupling image of the ECG signal. The proposed system uses the 2-D coupling image generated from three periods of the ECG signal as input data to the network whose design is based on a Convolutional Neural Network (CNN) that is specialized for image processing. Waveform of the 2-D coupling image which is the input data to the network cannot be visually confirmed and it has the advantage of being able to augment the QRS-complex which is a personal unique information. We confirm recognition performance of 99.2% from the experiment result for the proposed personal recognition system using MIT-BIH data.
Jin Su Kim; Sung Hyuck Kim; Sung Bum Pan. Personal recognition using convolutional neural network with ECG coupling image. Journal of Ambient Intelligence and Humanized Computing 2019, 11, 1923 -1932.
AMA StyleJin Su Kim, Sung Hyuck Kim, Sung Bum Pan. Personal recognition using convolutional neural network with ECG coupling image. Journal of Ambient Intelligence and Humanized Computing. 2019; 11 (5):1923-1932.
Chicago/Turabian StyleJin Su Kim; Sung Hyuck Kim; Sung Bum Pan. 2019. "Personal recognition using convolutional neural network with ECG coupling image." Journal of Ambient Intelligence and Humanized Computing 11, no. 5: 1923-1932.
As golf becomes more popular, interest in the analysis and correction of golf swing postures increases. Thus, various methods for analyzing the golf swing have been studied. Methods have recently been extended to 3D analysis systems, using motion capture technology beyond 2D analysis using cameras. The inertial sensor is one of the sensors used in 3D analysis. It is mainly used in the motion capture system, because it is cheap and has negligible space limitations. However, the motion capture system using the inertial sensor has some disadvantages in that it cannot identify body characteristics, such as joint length or the subject's initial direction, which are required to accurately analyze the swing data of the subject in golf, where the initial posture is important. This study proposes a system that obtains the same motion as the actual one by acquiring information on the motion of the subject using 15 inertial sensors and using information on the actual joint length and initial direction of the subject extracted from a depth camera. Experimental results show that the measurement error of information on the joint length and the foot stance of subjects extracted through the depth camera ranges from a minimum of 4.4% to a maximum of 6.94%. The acquired motion data from the proposed system confirms that the motion is the same as the one taken by the subject.
Syehyun Hwang; Kyeong‐Ri Ko; Sung Bum Pan. Motion data acquisition method for motion analysis in golf. Concurrency and Computation: Practice and Experience 2019, 33, 1 .
AMA StyleSyehyun Hwang, Kyeong‐Ri Ko, Sung Bum Pan. Motion data acquisition method for motion analysis in golf. Concurrency and Computation: Practice and Experience. 2019; 33 (2):1.
Chicago/Turabian StyleSyehyun Hwang; Kyeong‐Ri Ko; Sung Bum Pan. 2019. "Motion data acquisition method for motion analysis in golf." Concurrency and Computation: Practice and Experience 33, no. 2: 1.
The postmobile era will go beyond using individual smart devices and allow for user interaction by connecting various devices with sensing capabilities, such as smartphones, wearable devices, automobiles, and the Internet of Things. Wearable devices can continuously collect a variety of information on the users and their environment as the devices are worn in daily life. Because of this, real-time big data analysis technology is needed. This paper proposes a deep learning-based ensemble network model for improving the performance and overcoming the problems, which can occur on a single network. This model is designed so that the features produced by n number of single networks are combined and relearned. In addition, different parameter values are used on each single network, and the data used in the experiments are generated by the fiducial point method, which uses feature point detection, and the nonfiducial point method for periods of 1 sec and n sec. In the experiment results, in the case of fiducial point-based ECG signals, the ensemble network recognition performance shows a maximum of 0.8% higher accuracy than that of the single network. In the case of a 1 sec period nonfiducial point-based ECG signal, the ensemble network recognition performance is a minimum of 0.4% and a maximum of 1% higher than that of the single network. In the case of an n sec period, there is a maximum difference of 1.3%, and the proposed ensemble network shows better performance than the single network.
Min-Gu Kim; Sung Bum Pan. Deep Learning Based on 1-D Ensemble Networks Using ECG for Real-Time User Recognition. IEEE Transactions on Industrial Informatics 2019, 15, 5656 -5663.
AMA StyleMin-Gu Kim, Sung Bum Pan. Deep Learning Based on 1-D Ensemble Networks Using ECG for Real-Time User Recognition. IEEE Transactions on Industrial Informatics. 2019; 15 (10):5656-5663.
Chicago/Turabian StyleMin-Gu Kim; Sung Bum Pan. 2019. "Deep Learning Based on 1-D Ensemble Networks Using ECG for Real-Time User Recognition." IEEE Transactions on Industrial Informatics 15, no. 10: 5656-5663.
Studies have been actively conducted on biometrics technology applying electrocardiogram (ECG) signals, which are more robust against forgeries and alterations than fingerprint and face authentication. The ECG lead-I signals measured using ECG acquisition devices consist of one-dimensional (1D) data. Therefore, it has limitations with regard to feature extraction and data analysis. This paper proposes a user-recognition system that extracts multi-dimensional features through 2D resizing based on bi-cubic interpolation, which improves the calculation speed and preserves the original data values after converting the measured ECG into a spectrogram. An ECG measuring device was developed, and the ECGs were measured using the developed device. The proposed system consists of an ECG acquisition step, an ECG signal processing step, a segmentation step, a feature extraction step, and a classification step. For ECG signals, the CU-ECG dataset was created by acquiring ECG lead I signal data from 100 subjects in a relaxed state for a period of 160 s. For three sets of shuffle classes that applied the CU-ECG dataset, the average recognition performance was 93% for the existing algorithm and 88.9% for the parameter adjustment method. The average recognition performance of the proposed user recognition system showed a 0.33% improvement compared to the existing algorithm and a 4.43% improvement compared to the parameter adjustment method.
Gyu-Ho Choi; Eun-Sang Bak; Sung-Bum Pan. User Identification System Using 2D Resized Spectrogram Features of ECG. IEEE Access 2019, 7, 34862 -34873.
AMA StyleGyu-Ho Choi, Eun-Sang Bak, Sung-Bum Pan. User Identification System Using 2D Resized Spectrogram Features of ECG. IEEE Access. 2019; 7 (99):34862-34873.
Chicago/Turabian StyleGyu-Ho Choi; Eun-Sang Bak; Sung-Bum Pan. 2019. "User Identification System Using 2D Resized Spectrogram Features of ECG." IEEE Access 7, no. 99: 34862-34873.
This paper conducts a comparative analysis of deep models in biometrics using scalogram of electrocardiogram (ECG). A scalogram is the absolute value of the continuous wavelet transform coefficients of a signal. Since biometrics using ECG signals are sensitive to noise, studies have been conducted by transforming signals into a frequency domain that is efficient for analyzing noisy signals. By transforming the signal from the time domain to the frequency domain using the wavelet, the 1-D signal becomes a 2-D matrix, and it could be analyzed at multiresolution. However, this process makes signal analysis morphologically complex. This means that existing simple classifiers could perform poorly. We investigate the possibility of using the scalogram of ECG as input to deep convolutional neural networks of deep learning, which exhibit optimal performance for the classification of morphological imagery. When training data is small or hardware is insufficient for training, transfer learning can be used with pretrained deep models to reduce learning time, and classify it well enough. In this paper, AlexNet, GoogLeNet, and ResNet are considered as deep models of convolutional neural network. The experiments are performed on two databases for performance evaluation. Physikalisch-Technische Bundesanstalt (PTB)-ECG is a well-known database, while Chosun University (CU)-ECG is directly built for this study using the developed ECG sensor. The ResNet was 0.73%—0.27% higher than AlexNet or GoogLeNet on PTB-ECG—and the ResNet was 0.94%—0.12% higher than AlexNet or GoogLeNet on CU-ECG.
Yeong-Hyeon Byeon; Sung-Bum Pan; Keun-Chang Kwak. Intelligent Deep Models Based on Scalograms of Electrocardiogram Signals for Biometrics. Sensors 2019, 19, 935 .
AMA StyleYeong-Hyeon Byeon, Sung-Bum Pan, Keun-Chang Kwak. Intelligent Deep Models Based on Scalograms of Electrocardiogram Signals for Biometrics. Sensors. 2019; 19 (4):935.
Chicago/Turabian StyleYeong-Hyeon Byeon; Sung-Bum Pan; Keun-Chang Kwak. 2019. "Intelligent Deep Models Based on Scalograms of Electrocardiogram Signals for Biometrics." Sensors 19, no. 4: 935.