Multimodal 데이터에 대한 분류 에러 예측 기법

최의선,김재희,이철희,Choe, Ui-Seon,Kim, Jae-Hui,Lee, Cheol-Hui 대한전자공학회 2002 電子工學會論文誌-SP (Signal processing) Vol.39 No.2

본 논문에서는 multimodal 특성을 갖는 데이터에 대하여 패턴 분류 시 Bhattacharyya distance에 기반한 에러 예측 기법을 제안한다. 제안한 방법은 multimodal 데이터에 대하여 분류 에러와 Bhattacharyya distance를 각각 실험적으로 구하고 이 둘 사이의 관계를 유추하여 에러의 예측 가능성을 조사한다. 본 논문에서는 분류 에러 및 Bhattacharyya distance를 구하기 위하여 multimodal 데이터의 확률 밀도 함수를 정규 분포 특성을 갖는 부클래스들의 조합으로 추정한다. 원격 탐사 데이터를 이용하여 실험한 결과, multimodal 데이터의 분류 에러와 Bhattacharyya distance 사이에 밀접한 관련이 있음이 확인되었으며, Bhattacharyya distance를 이용한 에러 예측 가능성을 보여주었다. In this paper, we propose an error estimation method based on the Bhattacharyya distance for multimodal data. First, we try to find the empirical relationship between the classification error and the Bhattacharyya distance. Then, we investigate the possibility to derive the error estimation equation based on the Bhattacharyya distance for multimodal data. We assume that the distribution of multimodal data can be approximated as a mixture of several Gaussian distributions. Experimental results with remotely sensed data showed that there exist strong relationships between the Bhattacharyya distance and the classification error and that it is possible to predict the classification error using the Bhattacharyya distance for multimodal data.

소결온도에 따른 (1-x)Mg$TiO_3$-xSr$TiO_3$(x=0.02~0.08)세라믹스의 마이크로파 유전특성

최의선,이성갑,배선기,이영희,Choi, Eui-Sun,Lee, Sung-Gap,Bae, Seon-Gi,Lee, Young-Hie 한국전기전자재료학회 2000 전기전자재료학회논문지 Vol.13 No.12

The (1-x)MgTi $O_3$-xSrTi $O_3$(x=0.02~0.08) ceramics were fabricated by the conventional mixed oxide method. The sintering temperature and time were 125$0^{\circ}C$~1375$^{\circ}C$ and 2hours. The structure and microwave dielectric properties were investigated with sintering temperature and composition ratio. From the X-ray diffraction patterns, the cubic SrTi $O_3$and hexagonal MgTi $O_3$structures were coexisted in the (1-x)MgTi $O_3$-xSrTi $O_3$(x=0.02~0.08) ceramics. The dielectric constant($\varepsilon$$_{r}$) was increased and the temperature coefficient of resonant frequency($\tau$$_{f}$)was decreased with addition of SrTi $O_3$. The temperature coefficient of resonant frequency($\tau$$_{f}$) was gradually varied from negative value to positive value with increasing SrTi $O_3$. In the case of 0.96MgTi $O_3$-0.04SrTi $O_3$ceramics sintered at 130$0^{\circ}C$, the dielectric constant, quality factor and temperature coefficient of resonant frequency were 20.5, 5918(at 7.33GHz) and +10ppm/$^{\circ}C$, respectively.y.y.y.

소결온도와 Sr몰비에 따른 Mg$_{1-x}Sr_xTiO-3$ 세라믹스의 구조 및 마이크로파 유전특성

최의선,정장호,유기원,이영희,Choi, Eui-Sun,Chung, Jang-Ho,Ryu, Ki-Won,Lee, Young-Hie 대한전기학회 2001 전기학회논문지C Vol.49 No.5

The $Mg_{1-x}Sr_xTiO_3\;(x=0.02{\sim}0.08)$ ceramics were fabricated by the conventional mixed oxide method. The sintering temperature and time were $1250^{\circ}C{\sim}1350^{\circ}C$, 2hr., respectively. From the X-ray diffraction patterns, it was found that the perovskite $SrTiO_3$ and ilmenite $MgTiO_3$ structures were coexisted in the $Mg_{1-x}Sr_xTiO_3\;(x=0.02{\sim}0.08)$ ceramics. The dielectric constant( ${\epsilon}_r$) was increased with addition of $SrTiO_3$. The temperature coefficient of resonant frequency( ${\tau}_f$) was gradually varied from negative value to the positive value with increasing the $SrTiO_3$. The temperature coefficient of resonant frequency of the $Mg_{1-x}Sr_xTiO_3(x=0.036)$ ceramics was near zero, where the dielectric constant, quality factor, and ${\tau}_f$ were 20.65, 95120 and +1.3ppm/$^{\circ}C$, respectively.

$H_3BO_3$와 SnO 첨가에 따른 $0.6TiTe_3O_8-0.4MgTiO_3$ 세라믹스의 마이크로파 유전특성

최의선,이문기,류기원,이영희,김재식,Choi Eui-Sun,Lee Moon-Kee,Ryu Ki-Won,Lee Young-Hie,Kim Jae-Sik 대한전기학회 2005 전기학회논문지C Vol.54 No.4

In this study, the microwave dielectric properties of $0-6TiTe_3O_8-0.4MgTiO_3$ ceramic with $H_3BO_3$ and SnO were investigated to reduce the sintering time for the LTCC application. According to the X-ray diffraction patterns, both of $0-6TiTe_3O_8-0.4MgTiO_3$ ceramic with $H_3BO_3$ and SnO had the columbite structure of $TiTe_3O_8$ Phase, the ilmenite structure of $MgTiO_3$ phase. The density and dielectric constant of the $0-6TiTe_3O_8-0.4MgTiO_3$ ceramics with $H_3BO_3$ sintered at $830^{\circ}C$ for 1 hour were decreased but the quality factor was not changed with addition of $H_3BO_3$. Also the temperature coefficient of resonant frequency was not changed hardly. In the case of addition of SnO, the density and dielectric constant were increased but the quality factor was decreased and the temperature coefficient of resonant frequency was shifted to the negative(-) direction.

소결온도에 따른 (l-x)$TiTe_3O_{8}$-x$MgTiO_3$ 세라믹스의 마이크로파 유전 특성

최의선,김재식,이문기,류기원,이영희 대한전기학회 2004 전기학회논문지C Vol.53 No.9

In this study, the (l-x)$TiTe_3O_{8}$-x$MgTiO_3$ ceramics were investigated to obtain the improved dielectric properties of a high temperature stability and a sintering temperature of less than $900^{\circ}C$ which was necessary for the LTCC. According to the X-ray diffraction patterns of the (l-x)$TiTe_3O_{8}$-x$MgTiO_3$(x=0∼1) ceramics, the columbite structure of $TiTe_3O_{8}$ and ilmenite structure of $MgTiO_3$ were coexisted. Increasing the $MgTiO_3$ mole ratio(x), the density and dielectric constant were decreased and temperature coefficient of resonant frequency was moved to the negative direction and the quality factor was increased. In the case of the 0.6$TiTe_3O_{8}$-0.4$MgTiO_3$ ceramics sintered at $830^{\circ}C$ for 3hr., the microwave dielectric properties were $\varepsilon_{\gamma}$=29.3, Q${\times}$$f_{\gamma}$=39.600GHz and $\tau$$_{f}$=+9.3ppm/$^{\circ}C$.

소결온도에 따른 (1-x)TiTe3O8-xMgTiO3 세라믹스의 마이크로파 유전특성

최의선,김재식,이문기,류기원,이영희 대한전기학회 2004 전기학회논문지C Vol.53 No.9(C)

In this study, the (1-x)TiTe3O8-xMgTiO3 ceramics were investigated to obtain the improved dielectric properties of a high temperature stability and a sintering temperature of less than 900℃ which was necessary for the LTCC. According to the X-ray diffraction patterns of the (1-x)TiTe3O8-xMgTiO3(x=0~1) ceramics, the columbite structure of TiTe3O8 and ilmenite structure of MgTiO3 were coexisted. Increasing the MgTiO3 mole ratio(x), the density and dielectric constant were decreased and temperature coefficient of resonant frequency was moved to the negative direction and the quality factor was increased. In the case of the 0.6TiTe3O8-0.4MgTiO3 ceramics sintered at 830℃ for 3hr., the microwave dielectric properties were εr=29.3, Q×fr=39,600㎓ and τf=+9.3ppm/℃.

[隨筆] 기다림

崔義善 한국마케팅연구원 1970 마케팅 Vol.4 No.9

Multimodal 데이터에 대한 분류 에러 예측 기법

崔義善,김재희,이철희 대한전자공학회 2002 電子工學會論文誌-SP (Signal processing) Vol.39 No.3

In this paper, we propose an error estimation method based on the Bhattacharyya distance for multimodal data. First, we try to find the empirical relationship between the classification error and the Bhattacharyya distance. Then, we investigate the possibility to derive the error estimation equation based on the Bhattacharyya distance for multimodal data. We assume that the distribution of multimodal data can be approximated as a mixture of several Gaussian distributions. Experimental results with remotely sensed data showed that there exist strong relationships between the Bhattacharyya distance and the classification error and that it is possible to predict the classification error using the Bhattacharyya distance for multimodal data. 본 논문에서는 multimodal 특성을 갖는 데이터에 대하여 패턴 분류 시 Bhattacharyya distance에 기반한 에러 예측 기법을 제안한다. 제안한 방법은 multimodal 데이터에 대하여 분류 에러와 Bhattacharyya distance를 각각 실험적으로 구하고 이 둘 사이의 관계를 유추하여 에러의 예측 가능성을 조사한다. 본 논문에서는 분류 에러 및 Bhattacharyya distance를 구하기 위하여 multimodal 데이터의 확률 밀도 함수를 정규 분포 특성을 갖는 부클래스들의 조합으로 추정한다. 원격 탐사 데이터를 이용하여 실험한 결과, multimodal 데이터의 분류 에러와 Bhattacharyya distance 사이에 밀접한 관련이 있음이 확인되었으며, Bhattacharyya distance를 이용한 에러 예측 가능성을 보여주었다.

가우시안 분포의 다중클래스 데이터에 대한 최적 피춰추출 방법 ( Optimal feature extraction for normally distributed multiclass data )

최의선,이철희 대한전자공학회 1998 대한전자공학회 학술대회 Vol.1998 No.11

다중 클래스 데이터를 위한 분류오차 최소화기반 특징추출 기법

최의선,이철희,Choi, Eui-Sun,Lee, Chul-Hee 대한전자공학회 2000 電子工學會論文誌-SP (Signal processing) Vol.37 No.2

본 논문에서는 다중 클래스 데이터를 위한 특징 추출 방법을 최적화하는 기법을 제안한다 제안된 특징 추출 기법은 분류 오차에 기반한 방법으로 특징 공간(feature space)을 탐색하여 가우시안 최대우도 분류기 (Gaussian ML Classifier)의 분류오차(classification error)가 최소가 되도록 하는 특징벡터 집합을 구하는 방법이다 제안된 방법은 임의의 초기 특징벡터를 설정한 후 steepest descent 알고리즘을 적용하여 분류오차가 감소하는 방향으로 초기벡터를 갱신시킨다 본 논문에서는 순차탐색 및 전체탐색 두 가지의 방법을 제안하며 순차탐색은 추가로 특징벡터를 구하는 경우 이미 구해진 특징벡터를 포함하여 최소의 분류오차를 얻을 수 있는 특징벡터를 구한다 반면에 전체탐색 방법은 추가의 특징벡터를 구할 경우 새로운 초기 특징벡터 집합을 설정하여 이미 구해진 특징벡터를 포함하는 제약을 받지 않는다. 실험결과 제안된 두 가지 방법은 기존의 특징추출 방법보다 우수한 성능을 보여주고 있다. In this paper, we propose an optimizing feature extraction method for multiclass problems assuming normal distributions. Initially, We start with an arbitrary feature vector Assuming that the feature vector is used for classification, we compute the classification error Then we move the feature vector slightly in the direction so that classification error decreases most rapidly This can be done by taking gradient We propose two search methods, sequential search and global search In the sequential search, an additional feature vector is selected so that it provides the best accuracy along with the already chosen feature vectors In the global search, we are not constrained to use the chosen feature vectors Experimental results show that the proposed algorithm provides a favorable performance.