• 한국정보통신학회논문지
• /
• 제9권7호
• /
• pp.1497-1501
• /
• 2005

프로그램의 순차적인 실행 순서를 바꾸는 명령어를 분기 명령어라 하며, 분기는 마이크로프로세서의 파이프라인 정지를 일으켜 성능을 저하시키는 가장 큰 원인이 된다. 이에 따라 분기를 정확히 예측하여 다음 실행될 명령어를 제공한다면 마이크로프로세서의 자연스런 명령어의 실행 흐름은 끊어지지 않게 되고 이로써 논은 성능의 향상을 기대할 수 있게 된다. 분기 예측을 위해서는 분기 타겟 버퍼가 필수적이며, 분기 타겟 버퍼는 분기 예측 결과에 따라 다음에 실행할 명령어의 주소를 제공한다. 본 논문에서는 가상주소를 실제주소로 바꾸어 주는 TLB와 분기 타겟 버퍼가 각각 가지고 있는 태그 메모리를 함께 사용하는 구조를 제안한다. 이러한 공유 태그 구조의 이점은 2재의 태그 메모리를 하나로 공유함으로써 칩 면적의 감소를 꾀하고 더불어 분기 예측 속도를 향상시킬 수 있다는 점이다. 또한, 본 논문에서 제안된 구조는 주소로 사용되는 비트 수가 커지거나 여러 개의 명령어를 동시에 실행할 수 있는 구조에서 그 이점이 더욱 커지기 때문에 향후 개발되는 마이크로프로세서에서 유용하게 사용될 수 있을 것으로 기대된다.

• 한국산업융합학회 논문집
• /
• 제5권2호
• /
• pp.103-110
• /
• 2002

An artificial neural network is applied to the prediction of compressive strength, slump value of concrete. Standard mixed tables arc trained and estimated, and the results are compared with those of experiments. To consider the varieties of material properties, the standard mixed tables of two companies of Ready Mixed Concrete are used. And they are trained with the neural network. In this paper, standard back propagation network is used. For the arrangement on the approval of prediction of compressive strength and slump value, the standard compressive strength of 210, $240kgf/cm^2$ and target slump value of 12, 15cm are used because the amount of production of that range arc the most at ordinary companies. In results, in the prediction of compressive strength and slump value, the predicted values are converged well to those of standard mixed tables at the target error of 0.10, 0.05, 0.001 regardless of two companies.

• 대한임베디드공학회논문지
• /
• 제16권6호
• /
• pp.323-330
• /
• 2021

Visual object tracking is a challenging area of study in the field of computer vision due to many difficult problems, including a fast variation of target shape, occlusion, and arbitrary ground truth object designation. In this paper, we focus on the reinforced feature of the dynamic search area to get better performance than conventional discriminative model prediction trackers on the condition when the accuracy deteriorates since low feature discrimination. We propose a reinforced input feature method shown like the spotlight effect on the dynamic search area of the target tracking. This method can be used to improve performances for deep learning based discriminative model prediction tracker, also various types of trackers which are used to infer the center of the target based on the visual object tracking. The proposed method shows the improved tracking performance than the baseline trackers, achieving a relative gain of 38% quantitative improvement from 0.433 to 0.601 F-score at the visual object tracking evaluation.

• International Journal of Aeronautical and Space Sciences
• /
• 제3권1호
• /
• pp.39-49
• /
• 2002

This paper presents a robust tracker design method that is specific to the trajectories of target aircraft. This method assumes that representative trajectories of the target aircraft are available. The exact trajectories known to the tracker enables the incorporation of the exact data in the tracker design instead of the measurement data. An estimator is designed to have acceptable performance in tracking a finite number of different target trajectories with a capability to trade off the mean and maximum errors between the exact trajectories and the estimated or predicted trajectories. Constant estimator gains that minimize the cost functions related to the estimation or prediction error are computed off-line from an iterative algorithm. This tracker design method is applied to the longitudinal motion tracking of target aircraft.

• Journal of Information Technology Applications and Management
• /
• 제18권4호
• /
• pp.41-53
• /
• 2011

Trust prediction between users in social network based on the trust propagation assumes properties of transitivity and composability of trust propagation. But it has been hard to find studies which test on how those properties have been operated in real social network. This study aims to validate if the longer the distance of trust paths and the less the numbers of trust paths, the higher prediction error occurs using two real social network data set. As a result, the longer the distance of trust paths, we can find higher prediction error when predicting level of trust between source and target users. But we can not find decreasing trend of prediction error though the possible number of trust paths between source and target users increases.

• 한국안전학회지
• /
• 제22권3호
• /
• pp.28-33
• /
• 2007

Often Regular Measurement Target 32 Spots which are distributed at Seo-gu in Incheon Metropolitan and Odour Emission Target 100 Factories based on the task instruction of Ministry of Environment in Korea were selected by considering to atmosphere phenomena and regional characteristics etc. This paper aims at building the Decreasing Prediction System of Odour which is capable of comparing and examining the concentration distribution by odour compounds, the distribution maps of odour diffusion and the contribution degree of sphere of influence, which is discharged from these above spots and factories.

• 한국정밀공학회지
• /
• 제15권7호
• /
• pp.175-181
• /
• 1998

This paper describes realtime visual tracking system of moving object for three dimensional target using EKF(Extended Kalman Filter). We present a new realtime visual tracking using EKF algorithm and image prediction algorithm. We demonstrate the performance of these tracking algorithm through real experiment. The experimental results show the effectiveness of the EKF algorithm and image prediction algorithm for realtime tracking and estimated state value of filter, predicting the position of moving object to minimize an image processing area, and by reducing the effect by quantization noise of image.

• 한국항행학회논문지
• /
• 제13권2호
• /
• pp.208-213
• /
• 2009

본 연구는 모바일 환경에서 신경망 기법을 이용하여 서비스 호가 요구하는 대역폭의 크기를 예측하고, 목표 핸드오프호 손실확률 이하로 유지시키는 신경망 자원예측에 의한 적응 호 수락제어기법을 제안한다. 이 기법은 목표 핸드오프호의 손실확률을 설정하여 그 기준치 이상으로 핸드오프호의 손실확률이 발생하면 예약 대역폭의 양을 증가시켜부정확한 예측으로 인해 핸드오프호의 손실확률이 증가되는 것을 방지하여 핸드오프호의 GoS(Grade of Service)를 보장하기 위함이다. 제안한 신경망 자원예측과 목표 핸드오프호 손실확률에 기초한 적응 호수락제어기법의 성능을 기존의 호 수락제어기법과 비교하여 핸드오프호 손실확률을 기준치 이하로 유지할 수 있음을 보인다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제11권10호
• /
• pp.5006-5022
• /
• 2017

Effect and robust detection of targets in infrared images has crucial meaning for many applications, such as infrared guidance, early warning, and video surveillance. However, it is not an easy task due to the special characteristics of the infrared images, in which the background clutters are severe and the targets are weak. The recent literature demonstrates that sparse representation can help handle the detection problem, however, the detection performance should be improved. To this end, in this text, a hybrid method based on local sparse representation and contrast is proposed, which can effectively and robustly detect the infrared targets. First, a residual image is calculated based on local sparse representation for the original image, in which the target can be effectively highlighted. Then, a local contrast based method is adopted to compute the target prediction image, in which the background clutters can be highly suppressed. Subsequently, the residual image and the target prediction image are combined together adaptively so as to accurately and robustly locate the targets. Based on a set of comprehensive experiments, our algorithm has demonstrated better performance than other existing alternatives.

• 음성과학
• /
• 제10권4호
• /
• pp.7-21
• /
• 2003

Sums-of-Products models were built for segment duration prediction of spoken Korean. An experiment for the modelling was carried out to apply the results to Korean text-to-speech synthesis systems. 670 read sentences were analyzed. trained and tested for the construction of the duration models. Traditional sequential rule systems were extended to simple additive, multiplicative and additive-multiplicative models based on Sums-of-Products modelling. The parameters used in the modelling include the properties of the target segment and its neighbors and the target segment's position in the prosodic structure. Two optimisation strategies were used: the downhill simplex method and the simulated annealing method. The performance of the models was measured by the correlation coefficient and the root mean squared prediction error (RMSE) between actual and predicted duration in the test data. The best performance was obtained when the data was trained and tested by ' additive-multiplicative models. ' The correlation for the vowel duration prediction was 0.69 and the RMSE. 31.80 ms. while the correlation for the consonant duration prediction was 0.54 and the RMSE. 29.02 ms. The results were not good enough to be applied to the real-time text-to-speech systems. Further investigation of feature interactions is required for the better performance of the Sums-of-Products models.