• The Journal of the Korea Contents Association
• /
• v.13 no.4
• /
• pp.53-63
• /
• 2013

This paper presents an importance sampling(IS) embedded experimental frame(EF) design for efficient Monte Carlo (MC) simulation. To achieve IS principles, the proposed EF contains two embedded sub-models, which are classified into Importance Sampler(IS) and Bias Compensator(BC) models. The IS and BC models stand between the existing system model and EF, which leads to enhancement of model reusability. Furthermore, the proposed EF enables to achieve fast stochastic simulation as compared with the crude MC technique. From the abstract two case studies with the utilization of the proposed EF, we can gain interesting experimental results regarding remarkable enhancement of simulation performance. Finally, we expect that this work will serve various content areas for enhancing simulation performance, and besides, it will be utilized as a tool to understand and analyze social phenomena.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.10a
• /
• pp.204-207
• /
• 2015

이 연구는 공학 및 실험과학에서 활용되는 몬테 칼로 시뮬레이션 기법 중 하나인 무작위 행보 알고리즘의 성능 개선을 목표로 하였다. 이를 위해 무작위 행보 과정에서 난수 발생부와 행보 진행부를 분리하여 처리 시간을 단축하는 방안과, 문제 영역의 계산 규모를 2단계로 분할하여 시뮬레이션의 수렴 속도를 향상 시키는 방안을 제안한다. 또한 대규모 문제를 병렬처리 가능하도록 구현하고, 서로 다른 작업 분할 방식을 혼합하여 최적화를 수행 하였다. 순차 알고리즘만으로 실험한 결과 단순 구현방법과 비교해 실행시간과 에너지 소모량이 각각 18%의 성능향상을 얻었으며, 병렬 알고리즘을 8개의 노드(16코어)의 클러스터에서 실행했을 때 행 분할 방식의 성능이 블록 분할 방식보다 8% 빨라지는 것을 확인하였다.

• 한국방재학회:학술대회논문집
• /
• 2009.02b
• /
• pp.139.2-139.2
• /
• 2009

• The KIPS Transactions:PartB
• /
• v.8B no.3
• /
• pp.231-236
• /
• 2001

일반적인 다층 신경망에서 가중치의 갱신 알고리즘으로 사용하는 오류 역전과 방식은 가중치 갱신 결과를 고정된(fixed) 한 개의 값으로 결정한다. 이는 여러 갱신의 가능성을 오직 한 개의 값으로 고정하기 때문에 다양한 가능성들을 모두 수용하지 못하는 면이 있다. 하지만 모든 가능성을 확률적 분포로 표현하는 갱신 알고리즘을 도입하면 이런 문제는 해결된다. 이러한 알고리즘을 사용한 베이지안 신경망 모형(Bayesian Neural Networks Models)은 주어진 입력값(Input)에 대해 블랙 박스(Black-Box)와같은 신경망 구조의 각 층(Layer)을 거친 출력값(Out put)을 계산한다. 이 때 주어진 입력 데이터에 대한 결과의 예측값은 사후분포(posterior distribution)의 기댓값(mean)에 의해 계산할 수 있다. 주어진 사전분포(prior distribution)와 학습데이터에 의한 우도함수(likelihood functions)에 의해 계산한 사후확률의 함수는 매우 복잡한 구조를 가짐으로 기댓값의 적분계산에 대한 어려움이 발생한다. 따라서 수치해석적인 방법보다는 확률적 추정에 의한 근사 방법인 몬테 칼로 시뮬레이션을 이용할 수 있다. 이러한 방법으로서 Hybrid Monte Carlo 알고리즘은 좋은 결과를 제공하여준다(Neal 1996). 본 논문에서는 Hybrid Monte Carlo 알고리즘을 적용한 신경망이 기존의 CHAID, CART 그리고 QUEST와 같은 여러 가지 분류 알고리즘에 비해서 우수한 결과를 제공하는 것을 나타내고 있다.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.2 no.1
• /
• pp.90-100
• /
• 1999

유전자 알고리즘은 전통적인 등반 알고리즘을 이용하여 구하기 어려웠던 최적화 문제를 해결하기 위한 강인한(Robust) 탐색 기법이다. 특히 목적함수가 (1)여러 개의 국부 최대치를 가지는 경우, (2)수학적으로 표현이 불가능하거나 어려운 경우, (3)목적함수에 교란 항(disturbance term)이 섞여 있을 경우도 우수한 탐색 능력을 갖는 것으로 알려져 있다. 본 논문에서는 유전자 알고리즘을 이용하여 나타나는 다양한 해집합을 형성하는 개체군을 군집성 분석(cluster analysis)을 이용하여 군집화하고, 각 군집에 부여된 군집 적합도에 따라서 최적해를 구함으로써 단순 유전자 알고리즘에 의한 최적화보다 훨씬 향상된 탐색 알고리즘을 제안하였다. 반응표면의 형태가 정형화한 테스트 함수의 형태로 나타난다고 가정한 경우에 대하여 몬테 칼로 시뮬레이션을 통하여 본 알고리즘을 적용하여 평가하고 분석하였다.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.9 no.4
• /
• pp.5-14
• /
• 2006

The correct and opportune decision of reengaging the intercepted target is required in order to enhance the engagement performance of the surface to air missile systems that has the ability to defense or attack against various targets at the same time. The engagement efficiency and success of these systems will be largely enhanced by assigning quickly its system resources to the intercepted target and minimizing the waste of system resources for the target which is not able to attack any more. The kill-assessment algorithm has to be able to evaluate automatically whether various targets intercepted by missiles are killed or not on the basis of the reasonable confidence level. The definition of kill assessment is discussed and the kill assessment algorithm is designed reliably by using Kalman filter and a probability theory. Finally its performance is evaluated and analyzed by the Monte Carlo simulation.

• Journal of the Korean Ceramic Society
• /
• v.38 no.9
• /
• pp.846-852
• /
• 2001

A two-dimensional Monte Carlo simulation has been used to investigate the effect of the reaction temperature on the formation of the silicon carbide conversion layer near the surface of graphite substrate The carbothermal reduction of silica is the reaction mechanism of silicon carbide formation on graphite substrate by chemical vapor reaction methods. The chemical composition of silicon carbide conversion layer gradually changes from carbon to silicon carbide because gaseous reactants diffuse through micropores within graphite substrate and react with carbon at the surface of inner pores. The simulation was carried out under the condition of reaction temperature at 1900K, 2000K, 2100K and 2200K for 500MCS. It was found from the results of simulation that the thickness of silicon carbide conversion layer increases with reaction temperature.

• Journal of Internet Computing and Services
• /
• v.21 no.5
• /
• pp.1-7
• /
• 2020

Deep neural networks(DNN), which are used as approximation functions in reinforcement learning (RN), theoretically can be attributed to realistic results. In empirical benchmark works, time difference learning (TD) shows better results than Monte-Carlo learning (MC). However, among some previous works show that MC is better than TD when the reward is very rare or delayed. Also, another recent research shows when the information observed by the agent from the environment is partial on complex control works, it indicates that the MC prediction is superior to the TD-based methods. Most of these environments can be regarded as 5-step Q-learning or 20-step Q-learning, where the experiment continues without long roll-outs for alleviating reduce performance degradation. In other words, for networks with a noise, a representative network that is regardless of the controlled roll-outs, it is better to learn MC, which is robust to noisy rewards than TD, or almost identical to MC. These studies provide a break with that TD is better than MC. These recent research results show that the way combining MC and TD is better than the theoretical one. Therefore, in this study, based on the results shown in previous studies, we attempt to exploit a random balance with a mixture of TD and MC in RL without any complicated formulas by rewards used in those studies do. Compared to the DQN using the MC and TD random mixture and the well-known DQN using only the TD-based learning, we demonstrate that a well-performed TD learning are also granted special favor of the mixture of TD and MC through an experiments in OpenAI Gym.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.47 no.8
• /
• pp.574-581
• /
• 2019

This paper investigates the effects of thrust misalignments and offsets of the lateral center of gravity of a space launch vehicle on its guidance performance. Sensitivity analysis and Monte Carlo simulations are applied to analyze their effects by computing changes in orbit injection errors when including the error sources. To compensate their effects, the attitude controller including an integrator additionally and the Steering Misalignment Correction (SMC) routine of the Saturn V are considered, and then Monte Carlo simulations are performed to evaluate their performances.

• Journal of KSNVE
• /
• v.9 no.3
• /
• pp.509-516
• /
• 1999

The ESPI (Electronic Speckle Pattern Interferometry) algorithm has been simulated to calculate vibrational modes of a thin right-angled STS304 plate. The phase transformation of the reference wave of the ESPI is carried out only one time during vibration in order to clarify ESPI speckle patterns. Two dimensional vibrational modes are calculated from one ESPI pattern before vibration onset and two ESPI patterns during vibrations but with and without the phase transformation. The ESPI harmonic results are compared with those derived from the finite element method (FEM), and they agree very well. Additionally a phase unwrapping algorithm has been newly developed to derive a displacement map from an ESPI phase map.