http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
뇌파신호의 웨이블릿 변환 및 전처리에 따른 딥러닝 기반 발작 예측의 성능 비교
조용운(Yong Un Jo),오도창(Do Chang Oh) 대한전자공학회 2024 대한전자공학회 학술대회 Vol.2024 No.6
Epilepsy is a disease in which seizures occur irregularly. Sudden seizures during daily life can lead to serious accidents, and to prevent this, seizure prediction technology has been continuously researched. In this paper, we used the CHB-MIT database, including the period of seizures and their prodromes. We selected a portion of the database, applied three wavelet transforms, DWT, CWT, and TQWT, and classified them into ictal prodrome (preictal) and interictal period using a deep learning model. The results of three transform techniques are compared and a technique suitable for predicting patient seizures in real time is presented. Additionally, the size of the sliding window and the number of windows used were varied, and the prediction interval and predictable time were compared under various conditions. As a result, TQWT showed the best performance with 0.99 sensitivity, 0.94 f1 score, 0.09 FDR, and an average of 12 minutes in advance. For sliding window, using thirty windows of 30 seconds each showed the best performance.
근전도 기반의 Spider Chart와 딥러닝을 활용한 일상생활 잡기 손동작 분류
이성문,피승훈,한승호,조용운,오도창,Lee, Seong Mun,Pi, Sheung Hoon,Han, Seung Ho,Jo, Yong Un,Oh, Do Chang 대한의용생체공학회 2022 의공학회지 Vol.43 No.5
In this paper, we propose a pre-processing method that converts to Spider Chart image data for classification of gripping movement using EMG (electromyography) sensors and Convolution Neural Networks (CNN) deep learning. First, raw data for six hand gestures are extracted from five test subjects using an 8-channel armband and converted into Spider Chart data of octagonal shapes, which are divided into several sliding windows and are learned. In classifying six hand gestures, the classification performance is compared with the proposed pre-processing method and the existing methods. Deep learning was performed on the dataset by dividing 70% of the total into training, 15% as testing, and 15% as validation. For system performance evaluation, five cross-validations were applied by dividing 80% of the entire dataset by training and 20% by testing. The proposed method generates 97% and 94.54% in cross-validation and general tests, respectively, using the Spider Chart preprocessing, which was better results than the conventional methods.
음성인식과 화자검증을 통해 편리성과 보안성이 향상된 EMG 기반 능동의수 연구
김선홍(Seon-Hong Kim),김기승(Ki-Seung Kim),조용운(Yong-Un Jo),오도창(Do-Chang Oh) 대한전자공학회 2021 대한전자공학회 학술대회 Vol.2021 No.6
In this study, speaker verification and speech recognition technology are combined with an electronic prosthesis that performs basic movements based on EMG, and a new operation method is used to increase convenience and security. The speaker"s speech was trained using speech recognition and CNN on 4 hand gestures obtained from 10 subjects, resulting in an average of 97% accuracy for real-time speech data.
표면 근전도 신호와 딥러닝을 이용한 연속동작 인식 시스템 구현
이정민(Jeong Min Lee),송옥성(Ok Seong Song),이하진(Ha Jin Lee),조용운(Yong Un Jo),오도창(Do Chang Oh) 대한전자공학회 2024 대한전자공학회 학술대회 Vol.2024 No.6
In this study, a continuous motion recognition system was implemented by recognizing the EMG signals of the upper and lower arms. The signals are recognized using two products, the OY motion arm band and the myo arm band. Deep learning model CNN is used to distinguish various continuous motions as the recognized signals. When a test was conducted to distinguish the accuracy of the two actions through deep learning, the accuracy was at least 55% and up to 77%.