• 한국컴퓨터정보학회논문지
• /
• 제28권1호
• /
• pp.17-26
• /
• 2023

본 연구논문에서는 LSTM 기반의 학습 모델 적용과 그 효용성을 높일 수 있도록 몇 가지 평활 기법을 비교, 적용하고자 한다. 적용된 평활 기법은 Savitky-Golay, 지수 평활법, 가중치 이동 평균 등이다. 본 연구를 통해 비트코인 데이터에 LSTM모델 적용 시 보여준 결과 값보다 전처리 과정에서 적용된 Savitky-Golay 필터가 적용된 LSTM 알고리즘이 예측 성능에 유의미한 좋은 결과를 보였다. 예측 성능 결과를 확인하기 위해 비트코인 가격 예측에 따른 복잡 요인을 제거하는데 사용된 LSTM의 경우와 Savitzky-Golay LSTM 모델에 따른 학습 손실율과 검증 손실율을 비교하고 그 신뢰성을 높일 수 있도록 20회 평균값으로 실험하였다. 그 결과 (3.0556, 0.00005), (1.4659, 0.00002)의 값을 얻을 수 있었다. 결과적으로는 비트코인과 같은 암호화폐가 주식보다 더한 변동성을 가지는 만큼 데이터 전처리 과정에서 평활 기법(Savitzky-Golay)을 적용하여 잡음(Noise)을 제거하였으며, 전처리 후의 데이터는 LSTM 신경망 학습을 통해서 비트코인 예측률을 높이는데 가장 유의미한 결과를 얻을 수 있었다.

• 한국정밀공학회지
• /
• 제24권9호
• /
• pp.148-157
• /
• 2007

A new method which uses distance fields scheme and marching cube algorithm is proposed in order to get an accurate offset model of arbitrary shapes composed of triangular net. In the method, the space bounding the triangular net is divided into smaller cells. For the efficient calculation of distance fields, valid cells which will generate a portion of offset model are selected previously by the suggested detection algorithm. These valid cells are divided again into much smaller voxels which assure required accuracy. At each voxel distance fields are created by calculating the minimum distances between corner points of voxels and triangular net. After generating the whole distance fields, the offset surface were constructed by using the conventional marching cube algorithm together with mesh smoothing scheme. The effectiveness and validity of this new offset method was demonstrated by performing numerical experiments for the various types of triangular net.

• 한국멀티미디어학회논문지
• /
• 제19권2호
• /
• pp.233-239
• /
• 2016

Color enhancement basically deals with color manipulation in digital images. Recently, the technique has become widely used as a result of the increasing use of digital cameras. Retinex-based colorenhancement algorithms are a popular technique. In this paper, retinex with bilateral filtering is proposed to improve the quality of poorly illuminated images. Generally, it consists of three main steps: first, a retinex-based algorithm with color restoration; second, transformation mapping using histogram matching; and finally, smoothing the image using a bilateral filter. The experimental results demonstrate that the proposed method can successfully enhance image contrast while avoiding the halo effect and maintaining the color distribution in the image.

• 제어로봇시스템학회논문지
• /
• 제2권2호
• /
• pp.121-126
• /
• 1996

The hydrodynamic coefficients estimation (HCE) is important to design the autopilot and to predict the maneuverability of an underwater vehicle. In this paper, a system identification is proposed for an HCE of an underwater vehicle. First, we attempt to design the HCE algorithm which is insensitive to initial conditions and has good convergence, and which enables the estimation of the coefficents by using measured displacements only. Second, the sensor and measurement system which gauges the data from the full scale trials is constructed and the data smoothing algorithm is also designed to filter the noise due to irregular fluid flow without changing the data characteristics itself. Lastly the hydrodynamic coefficients are estimated by applying the measured data of full scale trials to the developed algorithm, and the estimated coefficients are verified by full scale trials.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 학술대회 논문집 정보 및 제어부문
• /
• pp.301-303
• /
• 2004

In this work, we develop on-line traffic prediction algorithm for real-time VBR traffic. There are a number of important issues: (i) The traffic prediction algorithm should exploit the stochastic characteristics of the underlying traffic and (ii) it should quickly adapt to structural changes in underlying traffic. GOP ARIMA model effectively addresses this issues and it is used as basis in our bandwidth prediction. Our prediction model deploy Kalman filter to incorporate the prediction error for the next prediction round. We examine the performance of GOP ARIMA based prediction with linear prediction with LMS and double exponential smoothing. The proposed prediction algorithm exhibits superior performam againt the rest.

• 대한전자공학회논문지
• /
• 제17권3호
• /
• pp.1-11
• /
• 1980

전자공학분야와 관련분야(일반력학, 물리 및 수학등) 사이의 용어의 차이를 해소하기 위한 노력을 계속하였고 통계학의 석학Box 씨와 Jenkins씨의 time series analysis의 입문을 위한 주변설명과 용어소개를 꾀하였다. 끝으로 Wiener의 filter와 Kalman-Bucy의 Algorithm을 설명하고 Hadamard를 위시한 변환기술의 유리점을 정리하여 보았다.

• 대한의용생체공학회:학술대회논문집
• /
• 대한의용생체공학회 1997년도 추계학술대회
• /
• pp.361-364
• /
• 1997

Bayesian MAP-EM approaches have been quite useful or tomographic reconstruction in that they can stabilize the instability of well-known ML-EM approaches, and can incorporate a priori information on the underlying emission object. However, MAP reconstruction algorithms with expressive priors often suffer from the optimization problem when their objective unctions are nonquadratic. In our previous work [1], we showed that the use of deterministic annealing method greatly reduces computational burden or optimization and provides a good solution or nonquadratic objective unctions. Here, we further investigate the convergence of the deterministic annealing algorithm; our experimental results show that, while the solutions obtained by a simple quenching algorithm depend on the initial conditions, the estimates converged via deterministic annealing algorithm are consistent under various initial conditions.

• 한국위성정보통신학회논문지
• /
• 제3권2호
• /
• pp.32-37
• /
• 2008

Global Navigation Satellite System (GNSS) is been developing in many countries. The satellite navigation system has the importance in economic and military fields. For utilizing satellite navigation system properly, the technology of GNSS Ground Station is needed. GNSS Ground Station monitors the signal of navigation satellite and analyzes navigation solution. This study deals with the navigation software for GNSS Ground Station. This paper will introduce the navigation solution algorithm for GNSS Ground Station. The navigation solution can be calculated by the code-carrier smoothing method, the Kalman-filter method, the least-square method, and the weight least square method. The performance of each navigation algorithm in this paper is presented.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2011년도 한국컴퓨터종합학술대회논문집 Vol.38 No.1(C)
• /
• pp.393-396
• /
• 2011

In this paper, we propose a constrained alternating least squares nonnegative matrix factorization algorithm (cALSNMF) to detect active brain regions from single subject's task-related fMRI data. In cALSNMF, we define a new cost function which considers the uncorrelation and noisy problems of fMRI data by adding decorrelation and smoothing constraints in original Euclidean distance cost function. We also generate a novel training procedure by modifying the update rules and combining with optimal brain surgeon (OBS) algorithm. The experimental results on visuomotor task fMRI data show that our cALSNMF fits fMRI data better than original ALSNMF in detecting task-related brain activation from single subject's fMRI data.

• 대한의용생체공학회:의공학회지
• /
• 제26권3호
• /
• pp.171-176
• /
• 2005

This paper proposes an automatic liver segmentation method using improved partial histogram threshold (PHT) algorithms. This method removes neighboring abdominal organs regardless of random pixel variation of contrast enhanced CT images. Adaptive multi-modal threshold is first performed to extract a region of interest (ROI). A left PHT (LPHT) algorithm is processed to remove the pancreas, spleen, and left kidney. Then a right PHT (RPHT) algorithm is performed for eliminating the right kidney from the ROI. Finally, binary morphological filtering is processed for removing of unnecessary objects and smoothing of the ROI boundary. Ten CT slices of six patients (60 slices) were selected to evaluate the proposed method. As evaluation measures, an average normalized area and area error rate were used. From the experimental results, the proposed automatic liver segmentation method has strong similarity performance as the MSM by medical Doctor.