• 항공우주시스템공학회지
• /
• 제10권1호
• /
• pp.8-14
• /
• 2016

The optimal estimation of a bearing only target tracking problem be achieved through the solution of the Fokker-Planck equation and the Bayesian update. Recently, a nonlinear filtering algorithm using a direct quadrature method of moments in which the associated Fokker-Planck equation can be propagated efficiently and accurately was proposed. Although this approach has demonstrated its promising in the field of nonlinear filtering in several examples, the "degeneracy" phenomenon, similar to that which exists in a typical particle filter, occasionally appears because only the weights are updated in the modified Bayesian rule in this algorithm. Therefore, in this paper to enhance the performance, a more stable measurement update process based upon the update equation in the Extended Kalman filters and a more accurate initialization and re-sampling strategy for weight and abscissas are proposed. Simulations are used to show the effectiveness of the proposed filter and the obtained results are promising.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권2호
• /
• pp.871-891
• /
• 2019

Visual tracking is a challenging task that needs learning an effective model to handle the changes of target appearance caused by factors such as pose variation, illumination change, occlusion and motion blur. In this paper, a novel tracking algorithm based on weighted local sub-space reconstruction error is presented. First, accounting for the appearance changes in the tracking process, a generative weight calculation method based on structural reconstruction error is proposed. Furthermore, a template update scheme of occlusion-aware is introduced, in which we reconstruct a new template instead of simply exploiting the best observation for template update. The effectiveness and feasibility of the proposed algorithm are verified by comparing it with some state-of-the-art algorithms quantitatively and qualitatively.

• 사물인터넷융복합논문지
• /
• 제8권5호
• /
• pp.33-39
• /
• 2022

모바일 컴퓨팅 분야에서 사용되는 저장장치는 저전력, 경량화, 내구성 등을 갖추어야 하며 사용자에 의해 생성되는 대용량 데이터를 효과적으로 저장 및 관리할 수 있어야 한다. 낸드 플래시 메모리는 모바일 컴퓨팅 분야에서 저장장치로 주로 사용되고 있다. 낸드 플래시 메모리는 구조적 특징 때문에 데이터 갱신요청 시 제자리 덮어쓰기가 불가능하여 데이터 갱신요청이 자주 발생하는 요청과 그렇지 않은 요청을 정확히 구분하여 각 블록에 저장 및 관리함으로써 해결할 수 있다. 이러한 데이터 갱신요청에 분류기법을 핫 데이터 식별 기법이라고 하며 현재 다양한 연구가 진행되었다. 본 논문은 더 정확한 핫 데이터 검증을 위해 카운팅 필터를 사용하여 데이터 갱신요청 발생을 연속적으로 기록하고 또한 특정 시간 동안 요청된 갱신요청이 얼마나 자주 발생하는지를 고려하여 핫 데이터를 검증한다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 ITC-CSCC -2
• /
• pp.1041-1044
• /
• 2000

The enhancement to the back-propagation algorithm presented in this paper has resulted from the need to extract sparsely connected networks from networks employing product terms. The enhancement works in conjunction with the back-propagation weight update process, so that the actions of weight zeroing and weight stimulation enhance each other. It is shown that the error measure, can also be interpreted as rate of weight change (as opposed to ${\Delta}W_{ij}$), and consequently used to determine when weights have reached a stable state. Weights judged to be stable are then compared to a zero weight threshold. Should they fall below this threshold, then the weight in question is zeroed. Simulation of such a system is shown to return improved learning rates and reduce network connection requirements, with respect to the optimal network solution, trained using the normal back-propagation algorithm for Multi-Layer Perceptron (MLP), Higher Order Neural Network (HONN) and Sigma-Pi networks.

• International journal of advanced smart convergence
• /
• 제8권4호
• /
• pp.68-74
• /
• 2019

Along with the deeper architecture in the deep learning approaches, the need for the data becomes very big. In the real problem, to get huge data in some disciplines is very costly. Therefore, learning on limited data in the recent years turns to be a very appealing area. Meta-learning offers a new perspective to learn a model with this limitation. A state-of-the-art model that is made using a meta-learning framework, Meta-SGD, is proposed with a key idea of learning a hyperparameter or a learning rate of the fast adaptation stage in the outer update. However, this learning rate usually is set to be very small. In consequence, the objective function of SGD will give a little improvement to our weight parameters. In other words, the prior is being a key value of getting a good adaptation. As a goal of meta-learning approaches, learning using a single gradient step in the inner update may lead to a bad performance. Especially if the prior that we use is far from the expected one, or it works in the opposite way that it is very effective to adapt the model. By this reason, we propose to add a weight term to decrease, or increase in some conditions, the effect of this prior. The experiment on few-shot learning shows that emphasizing or weakening the prior can give better performance than using its original value.

• The Journal of the Acoustical Society of Korea
• /
• 제13권1E호
• /
• pp.25-36
• /
• 1994

본 논문에서는 외부로부터 제공되는 학습데이타에 신경회로망의 자기적응화(self-adaptation)를 이룩하기 위한 접근론이 기술된다. 이러한 문제점은 신경회로망의 학습이론, 즉 현재의 학습 데이터에 적절한 신경회로망이 가중치 벡터들(weight vectors)의 개선 방법론에 기인된다. 이들에 관련된 문제점들의 이론적 검토와 아울러 신경회로망의 학습에 대한 근본적인 요소들이 재조명된다. 현재 가장 널리 이용되고 있는 후방 전달(back-propagation) 학습법과 비교함으로써, 본 연구에서 제안된 자기적응 학습법의 유용성과 우위성을 컴퓨터 모의시험 결과로 입증하게 된다.

• 한국군사과학기술학회지
• /
• 제19권4호
• /
• pp.526-535
• /
• 2016

In this study, the estimating weight data of eight leading ships of Korea navy were processed by comparing with the final weight data of their incline test results statistically, and are analyzed as follows; The maximum criteria of the weight margins of Korea naval ship in the preliminary and detailed design phases seem to be appropriate values, but the minimum criteria would be advisable to update more reasonable values. And, the limitation that the acceptable deviation of lightship weight should be 10 % of the lightship weight on a naval ship's ROC(Required Operational Capabilities) is recommended to be prohibited, because it comes from the weight estimation which has considerable uncertainty and it may also drop off design flexibility. Finally, the SWBS(Ship Work Breakdown System) groups which have larger deviation values in the estimating weight of naval ships are necessary to improve their accuracies, and to upgrade their weight database continuously.

• 한국멀티미디어학회논문지
• /
• 제10권8호
• /
• pp.1004-1016
• /
• 2007

본 논문에서는 기준신호를 획득하기 어려운 환경에서 환경소음이 정상적인 특성을 가질 경우 음성을 향상시킬 수 있는 가중치 갱신제어 적응소음제거기를 제안하였다. 일반적인 적응소음제거기의 경우 소음만의 기준신호를 획득하여야 한다. 그러나 다수의 기기에 의한 복합적인 소음과 작업자에 의한 음성이 혼합되는 공장 환경에서는 소음발생원들로 부터 순수한 소음신호를 획득하기가 어렵다. 따라서 기준신호를 이용할 수 없기 때문에 이러한 환경에서는 기존의 적응잡음제거기를 사용하기가 어렵다. 제안한 방법에서는 입력신호를 임의의 상수로 하고 기준신호에 마이크로폰의 신호를 입력한다. 그런 다음 음성이 없는 구간에서 적응필터의 가중치를 갱신하여 소음을 제거하고 음성이 발생한 구간에서는 가중치를 고정하여 소음이 제거된 변형 음성신호를 획득한다. 그리고 변형 음성신호를 복원 필터링하여 음성신호를 출력한다. 이것은 다수의 공장소음이 정상적이고 짧은 대화구간에서 소음이 변하지 않는 점을 고려하였다. 실험의 결과 제안한 소음제거기가 공장소음을 효과적으로 제거할 수 있었고 신호 대 잡음비 면에서도 우수함을 확인하였다.

• 한국해양공학회지
• /
• 제27권5호
• /
• pp.36-42
• /
• 2013

We present the optimum design for the cross-sectional(sizing) and shape optimization of truss structures with natural frequency constraints. The optimum design method used in this paper employs continuous design variables and the Harmony Search Algorithm(HSA). HSA is a meta-heuristic search method for global optimization problems. In this paper, HSA uses the method of random number selection in an update process, along with penalty parameters, to construct the initial harmony memory in order to improve the fitness in the initial and update processes. In examples, 10-bar and 72-bar trusses are optimized for sizing, and 37-bar bridge type truss and 52-bar(like dome) for sizing and shape. Four typical truss optimization examples are employed to demonstrate the availability of HSA for finding the minimum weight optimum truss with multiple natural frequency constraints.

• 한국컴퓨터정보학회논문지
• /
• 제22권12호
• /
• pp.49-54
• /
• 2017

Pathfinding for pedestrians provided by various navigation programs is based on a shortest path search algorithm. There is no big difference in their guide results, which makes the path quality more important. Multiple criteria should be included in the search cost to calculate the path quality, which is called a multi-criteria pathfinding. In this paper we propose a user adaptive pathfinding algorithm in which the cost function for a multi-criteria pathfinding is defined as a weighted sum of multiple criteria and the weights are learned automatically by Perceptron learning. Weight learning is implemented in two ways: short-term weight learning that reflects weight changes in real time as the user moves and long-term weight learning that updates the weights by the average value of the entire path after completing the movement. We use the weight update method with momentum for long-term weight learning, so that learning speed is improved and the learned weight can be stabilized. The proposed method is implemented as an app and is applied to various movement situations. The results show that customized pathfinding based on user preference can be obtained.