http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
- 中文 을 입력하시려면 zhongwen을 입력하시고 space를누르시면됩니다.
- 北京 을 입력하시려면 beijing을 입력하시고 space를 누르시면 됩니다.
RISS 인기검색어
- 검색키워드
- 저자 : ( Vikneswaran Vijean ) (검색결과 3 건)
- 무료
- 기관 내 무료
- 유료
-
Discrimination of False Arrhythmia Alarms in Intensive Care Unit (ICU) using ECG Signals
( Vikneswaran Vijean ),( Gunashareene R. Pavinthiran ),( M. Hariharan ) 한국감성과학회 2017 한국감성과학회 국제학술대회(ICES) Vol.2017 No.-
False arrhythmia alarm is the insignificant alarm caused by the existing monitoring instruments that are with an algorithm that sets off alarms in high decibel sounds. The paper proposes an algorithm that is capable of distinguishing the arrhythmia into true and false events accurately. The electrocardiogram (ECG) signals were pre-processed to remove the common noise exists such as the baseline wander, trends and the pacemaker spiking. Wavelet Transform method with Daubechies 4 (Db 4) was used to extract Shannon entropy, Renyi entropy, Tsallis entropy and Log-Energy entropy at the specified level of decomposition of the Db4.The other feature type that would be extracted is the morphological features based on the temporal information of the ECG signals namely the heart rate, RR- interval and the R peak amplitude which was obtained using the Pan and Tompkins algorithm. The features that selected through Welch T-Test were further processed for data synchronization using ADASYN and for dimensionality reduction using Principle Component Analysis (PCA). The features then were classified using Fuzzy K-Nearest Neighbor (FKNN), Random Forest classifier, and AdaBoost classifier. The highest accuracy attained was 85.14%, 81.94% and 80.77% for the respective classifiers on the principle components (PCA) of synchronized (ADASYN) feature of the combined statistical and morphological type of feature.
-
( Rokiah Abdullah ),( M. Hariharan ),( Vikneswaran Vijean ),( Farah Nazlia Che Kassim ),( Zulkapli Abdullah ) 한국감성과학회 2017 한국감성과학회 국제학술대회(ICES) Vol.2017 No.-
Malaysia mainly consists of people from three ethnic groups (Malay, Chinese and Indian) and the official and national language of Malaysia is Malay. Malaysian standard English is a form English used and spoken in in Malaysia. Due to the heavy influence of Malay and lack of use of English, Malaysians English accent of Malaysian is not similar to standard English language. So, it is difficult use most commercially available speaker and accent recognition system for Malaysian speaker and accent recognition. Therefore, it is necessary to develop a system to recognize individuals and their accents using the spoken utterances (Malaysian English). Spoken utterances (English digits (0~9) and Malay words) are recorded using the subjects including three ethnic groups (Malay, Chinese and Indian) of Malaysia. Database of speaker and accent are developed using. In this paper, wavelet packet transform based feature extraction method is proposed to extract salient features from the recorded spoken utterances. Support vector machine (SVM) is used to recognize individuals and their accent. The highest accuracy of 92.90% was obtained for speaker identification and 94.89% for accent identification using Malay words whereas the highest accuracy of 91.18% was obtained for speaker identification and 94.17% for accent recognition using Malay digit.
-
Higher Order Spectra based Features for Infant Cry Signal Classification
( M. Hariharan ),( Yogesh C. K. R. Sindhu ),( Vikneswaran Vijean ),( Haniza Yazid ),( Thiyagar Nadarajaw ),( Kemal Polat ),( Sazali Yaacob ) 한국감성과학회 2017 한국감성과학회 국제학술대회(ICES) Vol.2017 No.-
Crying is an early communication medium for infants. The cry analysis provides an opportunity to assess the physical and pathological status of infants. Several short-term cepstral/spectral features were extracted from the recorded cry signals to detect the reason for crying (hunger, pain, sick condition etc). In this work, Higher order spectra (HOS) based features were proposed to study its efficacy in better representation of cry signals. Statistical features from Bi-spectral plots were derived. Two well-known short-term feature sets known as Mel-frequency cepstral coefficients (MFCCs) and Linear predictive coding based cepstral coefficients (LPCs, Linear predictive cepstral coefficients- LPCCs and weighted LPCCs) were also used. Cry signals from 2 different databases were utilized. Several experiments of twoclass and multi-class classification of cry signals were conducted and the results were reported to demonstrate the effectiveness of the proposed method. The experimental results indicate proposed combination of HOS based features with standard short-term cepstral features help to achieve promosing infant cry classification accuracy.
연관 검색어 추천
이 검색어로 많이 본 자료
활용도 높은 자료
