http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
나승유 ( Na Seung Yu ),신대정 ( Sin Dae Jeong ) 한국센서학회 2004 센서학회지 Vol.13 No.1
N/A Due to various types of errors added to dynamic weight measurement data, proper methods to reduce measurement errors are required to produce reliable weights. To cope with parasitic types of errors in real systems, information provided by the various sensors is utilized and combined in such a way to reduce the measurement errors of load cells. In addition to four channels of load cells from a trailer, an accelerometer is used to obtain the information to compensate the error induced from vertical movement of the vehicle due to the variation of ground level. A model trailer system is run to verify the effectiveness of the proposed method to reduce noise of dynamic weight measurements. Experiments show that the processed error magnitudes of less than 20 g can be obtained for 10 Kg experimental loads.
최적화된 관측 신뢰도와 변형된 HMM 디코더를 이용한 잡음에 강인한 화자식별 시스템
김진영,나승유,Kim, Jin-Young,Na, Seung-Yu 대한음성학회 2007 말소리 Vol.64 No.-
Speech signal is distorted by channel characteristics or additive noise and then the performances of speaker or speech recognition are severely degraded. To cope with the noise problem, we propose a modified HMM decoder algorithm using SNR-based observation confidence, which was successfully applied for GMM in speaker identification task. The modification is done by weighting observation probabilities with reliability values obtained from SNR. Also, we apply PSO (particle swarm optimization) method to the confidence function for maximizing the speaker identification performance. To evaluate our proposed method, we used the ETRI database for speaker recognition. The experimental results showed that the performance was definitely enhanced with the modified HMM decoder algorithm.
논문 : 화자인식 성능 향상을 위한 Particle Swarm 방법 기반 Sigmoid 형 멤버쉽 함수 최적화
김진영 ( Jin Young Kim ),나승유 ( Seung Yu Na ),최승호 ( Seung Ho Choi ) 전남대학교 전자통신기술연구소 2006 전자통신기술논문지 Vol.9 No.1
화자인식기의 성능은 잡음환경 하에서 크게 저하된다. 본 논문에서는 잡음에 의해 오염된 관측 값의 멤버쉽 값을 sigmoid 함수로 표현하고, particle swarm 최적화 방법을 이용하여, 최적화하는 방법을 제안하였다. 관측 멤버쉽 값은 각 화자 모델의 확률 계산 시 가중값으로 이용되며, 화자인식 실험결과 식별률이 크게 향상됨을 확인하였다. The performance of speaker identifier is severly degraded in noisy environments. In this paper we formulate observation membership with sigmoid function. Also, we optimize the parameters of sigmoid function with particle swarm optimization method. In test stage, the observation membership is used in calculating probability as weighting value. The experimental results show that the proposed method works will in speaker identification.