• Journal of Internet Computing and Services
• /
• v.17 no.1
• /
• pp.1-6
• /
• 2016

The biased zero-error probability and its related algorithms require heavy computational burden related with some summation operations at each iteration time. In this paper, a recursive approach to the biased zero-error probability and related algorithms are proposed, and compared in the simulation environment of shallow water communication channels with ambient noise of biased Gaussian and impulsive noise. The proposed recursive method has significantly reduced computational burden regardless of sample size, contrast to the original MBZEP algorithm with computational complexity proportional to sample size. With this computational efficiency the proposed algorithm, compared with the block-processing method, shows the equivalent robustness to multipath fading, biased Gaussian and impulsive noise.

• Journal of Internet Computing and Services
• /
• v.14 no.1
• /
• pp.9-14
• /
• 2013

The criterion of zero-error probability provides a limitation on error probability functions being used for adaptive systems when the error samples are shifted by the influence of DC-bias noise. In this paper, we employ a bias term in the error distribution and propose a new criterion of the biased zero-error probability with error being zero. Also, by maximizing the proposed criterion on expanded filter structures, a supervised adaptive algorithm has been derived. From the simulation results of supervised equalization, the algorithm based on the proposed criterion yielded zero-centered and highly concentrated error samples without disturbance in the environments of strong impulsive and DC-bias noise.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.14 no.3
• /
• pp.1416-1421
• /
• 2013

In this paper, the Euclidean distance between the probability density functions (PDFs) for biased errors and a Dirac-delta function located at zero on the error axis is proposed as a new performance criterion for adaptive systems in non-Gaussian noise environments. Also, based on the proposed performance criterion, a supervised adaptive algorithm is derived and applied to adaptive equalization in the shallow-water communication channel distorted by severe multipath fading, impulsive and DC-bias noise. The simulation results compared with the performance of the existing MEDE algorithm show that the proposed algorithm yields over 5 dB of MSE enhancement and the capability of relocating the mean of the error PDF to zero on the error axis.

• Proceedings of the Korean Information Science Society Conference
• /
• 2005.07b
• /
• pp.619-621
• /
• 2005

은닉 마코프 모델은 이산 동역학을 표현할 수 있는 확률 모형이다. 우도 함수 최적화를 수행하는 전통적인 Baum-Welch 학습 알고리즘은 국소해로 수령하기 쉬우며, 우도함수의 특성상 복잡한 모델을 선호하는 바이어스가 존재한다. 베이지안 프레임워크에서는 파라미터를 랜덤 변수로 보고 이에 대한 사후 확률 분포를 추정하여 이 문제를 해결할 수 있다. 본 논문에서는 베이지안 추정을 위한 결정론적 근사화 기법인 변분 베이지안 방법을 이용, 출력 노드에 가우시안 혼합 노드를 지니는 일반화된 HMM의 추론 방법을 유도한다. 인공 데이터에 대한 실험을 통해, 본 방법이 효과적인 HMM 학습을 수행할 수 있음을 보인다.

• Annual Conference of KIPS
• /
• 2005.05a
• /
• pp.431-434
• /
• 2005

근원데이터나, 이원데이터를 이용한 문제를 해결하기 위해서는 많은 경우에 완전 해를 갖는 문제로 변형시키기 위해 정규화할 필요성이 있다. 본 논문에서는 이러한 정규화 인수를 찾는 문제를 기존의 GCV, L-Curve, 그리고 이원데이터를 RBF 신경회로망에 적용시킨 커널 학습법에 대한 각각의 성능을 비교실험을 통해 고찰한다. 이때 커널을 이용한 학습법의 성능을 향상하기 위해, 전체학습과 성능의 제한적 비례관계라는 설정아래, 각각의 학습에 따라 능동적으로 변화하는 동적모멘텀의 도입을 제안한다. 끝으로 제안된 동적모멘텀이 분류문제의 표준인 Iris 데이터, Singular 시스템의 대표적 모델인 가우시안 데이터, 그리고 마지막으로 1차원 이미지 복구문제인 Shaw데이터를 이용한 각각의 실험에서 분류문제와 회계문제 양쪽 모두에 있어 기존의 GCV, L-Curve와 동등하거나 우수한 성능이 있음을 보인다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2009.10a
• /
• pp.917-919
• /
• 2009

본 연구에서는 산화막 두께에 따른 20nm 이하 MOSFET의 전류-전압 특성 곡선 분석하였다. 산화물 내의 등가 포획 전하는 가우시안 함수를 사용하였다. 채널의 길이가 20nm 이하인 LDD MOSFET를 설계하여 사용하였고, 소자를 시뮬레이션 하기 위하여 실리콘 공정 디바이스 시뮬레이터인 MicroTec의 SemSim을 사용하였다. SemSim은 디바이스 시뮬레이터로써 입력 바이어스에 의해 공정 시뮬레이션인 SiDif와 디바이스 조립인 MergIC에 의해 소자를 시뮬레이션 한다. 산화막의 두께를 2nm, 3nm, 4nm로 시뮬레이션 한 결과 산화막의 두께가 얇아짐에 따라 드레인에 흐르는 전류가 증가함을 알 수 있었다.

• Journal of Advanced Navigation Technology
• /
• v.3 no.2
• /
• pp.97-107
• /
• 1999

Passive sensors such as Laser Range Finder(LRF) and Forward Looking Infrared(FLIR) camera frequently used for tracking aircraft landing produce the measurements of elevation angle contaminated by large noise due to the exhaust plume disturbance. This results in poor tracking performance if the extended Kalman filter is used for estimation of the range and elevation which are corrupted by the non-Gaussian noise such as plume disturbance. In this paper, an adaptive estimation filter and the extended Kalman filter is combined to produce a combination-type filter. In this approach the adaptive filter is used for the plume-type disturbance noise and the extended Kalman filter is utilized for the measurement of Gaussian type. The proposed combination filter is effective for the trajectory estimation of landing aircraft under the influence of unknown bias and numerical simulations illustrate the performance of the proposed filter.

• Journal of KIISE:Software and Applications
• /
• v.32 no.11
• /
• pp.1021-1028
• /
• 2005

The general solution for classification and regression problems can be found by matching and modifying matrices with the information in real world and then these matrices are teaming in neural networks. This paper treats primary space as a real world, and dual space that Primary space matches matrices using kernel. In practical study, there are two kinds of problems, complete system which can get an answer using inverse matrix and ill-posed system or singular system which cannot get an answer directly from inverse of the given matrix. Further more the problems are often given by the latter condition; therefore, it is necessary to find regularization parameter to change ill-posed or singular problems into complete system. This paper compares each performance under both classification and regression problems among GCV, L-Curve, which are well known for getting regularization parameter, and kernel methods. Both GCV and L-Curve have excellent performance to get regularization parameters, and the performances are similar although they show little bit different results from the different condition of problems. However, these methods are two-step solution because both have to calculate the regularization parameters to solve given problems, and then those problems can be applied to other solving methods. Compared with UV and L-Curve, kernel methods are one-step solution which is simultaneously teaming a regularization parameter within the teaming process of pattern weights. This paper also suggests dynamic momentum which is leaning under the limited proportional condition between learning epoch and the performance of given problems to increase performance and precision for regularization. Finally, this paper shows the results that suggested solution can get better or equivalent results compared with GCV and L-Curve through the experiments using Iris data which are used to consider standard data in classification, Gaussian data which are typical data for singular system, and Shaw data which is an one-dimension image restoration problems.