• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제24권4호
• /
• pp.331-350
• /
• 2020

Exploding gradient is a widely known problem in training recurrent neural networks. The explosion problem has often been coped with cutting off the gradient norm by some fixed value. However, this strategy, commonly referred to norm clipping, is an ad hoc approach to attenuate the explosion. In this research, we opt to view the problem from a different perspective, the discrete-time optimal control with infinite horizon for a better understanding of the problem. Through this perspective, we fathom the region at which gradient explosion occurs. Based on the analysis, we introduce a gradient-explosion-free algorithm that keeps the training process away from the region. Numerical tests show that this algorithm is at least three times faster than the clipping strategy.

• 융합보안논문지
• /
• 제18권3호
• /
• pp.3-9
• /
• 2018

DDPG(Deep Deterministic Policy Gradient)알고리즘은 인공신경망과 강화학습을 사용하여 학습하는 알고리즘이다. 최근 많은 연구가 이루어지고 있는 강화학습과 관련된 연구 중에서도 DDPG 알고리즘은 오프폴리시로 학습하기 때문에 잘못된 행동이 누적되어 학습에 영향을 미치는 경우를 방지하는 장점이 있다. 본 연구에서는 DDPG 알고리즘을 응용하여 자전거를 자율주행 하도록 제어하는 실험을 진행하였다. 다양한 환경을 설정하여 시뮬레이션을 진행하였고 실험을 통해서 사용된 방법이 시뮬레이션 상에서 안정적으로 동작함을 보였다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1998년도 하계학술대회 논문집 B
• /
• pp.515-518
• /
• 1998

This paper proposes a new hybrid algorithm for auto-tuning fuzzy controllers improving the performance. The presented algorithm estimates automatically the optimal values of membership functions, fuzzy rules, and scaling factors for fuzzy controllers, using a genetic-MGM algorithm. The object of the proposed algorithm is to promote search efficiency by a genetic and modified gradient optimization techniques. The proposed genetic and MGM algorithm is based on both the standard genetic algorithm and a gradient method. If a maximum point don't be changed around an optimal value at the end of performance during given generation, the genetic-MGM algorithm searches for an optimal value using the initial value which has maximum point by converting the genetic algorithms into the MGM(Modified Gradient Method) algorithms that reduced the number of variables. Using this algorithm is not only that the computing time is faster than genetic algorithm as reducing the number of variables, but also that can overcome the disadvantage of genetic algorithms. Simulation results verify the validity of the presented method.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 1994년도 봄 학술발표회 논문집
• /
• pp.38-45
• /
• 1994

The general conceptual constitution of structural optimization is formulated. The algorithm using the gradient projection method and design sensitivity analysis is discussed. Examples of minimum-weight design for six-story steel plane frame are taken to illustrate the application of this algorithm. The advantages of this algorithm such as marginal cost and design sensitivity analysis as well as system analysis are explained.

• 산업기술연구
• /
• 제24권B호
• /
• pp.143-148
• /
• 2004

We present a fast algorithm for CDMA (code division multiple access) multiuser detection using the gradient guided search. The fast algorithm calculates the maximum likelihood (ML) metric so efficiently that it needs only O(K) additions in the presence of K users once some initialization is completed. The computational advantages of the fast algorithm over the conventional method are more noticeable as more iterations are required to obtain a suboptimal solution as in the initialization with matched filters.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2007년도 하계종합학술대회 논문집
• /
• pp.389-390
• /
• 2007

In this paper, we propose a modified watershed algorithm to obtain exact edge of region. The proposed method adjusts priority at zero-crossing point of gradient in order to make the point of region decision time postponed. We compare the proposed method with a previous method and prove that this method can extract more correct edge of region.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1996년도 하계학술대회 논문집 B
• /
• pp.1371-1373
• /
• 1996

A clustering algorithms with dynamic adjustment of learning rate for GBFCM(Gradient Based FCM) is proposed in this paper. This algorithm combines two idea of dynamic K-means algorithms and GBFCM : learning rate variation with entropy concept and continuous membership grade. To evaluate dynamic GBFCM, we made comparisons with Kohonen's Self-Organizing Map over several tutorial examples and image compression. The results show that DGBFCM(Dynamic GBFCM) gives superior performance over Kohonen's algorithm in terms of signal-to-noise.

• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 1998년도 The Third Asian Fuzzy Systems Symposium
• /
• pp.740-749
• /
• 1998

This paper proposes compression of image data using neural networks based on conjugate gradient method and dynamic tunneling system. The conjugate gradient method is applied for high speed optimization .The dynamic tunneling algorithms, which is the deterministic method with tunneling phenomenon, is applied for global optimization. Converging to the local minima by using the conjugate gradient method, the new initial point for escaping the local minima is estimated by dynamic tunneling system. The proposed method has been applied the image data compression of 12 ${\times}$12 pixels. The simulation results shows the proposed networks has better learning performance , in comparison with that using the conventional BP as learning algorithm.

• 대한교통학회지
• /
• 제18권5호
• /
• pp.99-107
• /
• 2000

경로기반 통행 배정은 실시간 통행 배정에서 이용되는 경로기반 해를 제공할 수 있기 때문에 첨단 교통 체계(ITS)의 실시간 교통 제어 및 교통 안내 등에 유용하게 이용될 수 있다. 많이 사용되고 있는 경로기반 통행배정 알고리즘의 하나인 Gradient Projection(GP) 알고리즘은 일반적으로 최적해 근처로는 빠른 접근 속도를 보이나. 일단 최적해에 근접하면 수렴 속도가 다소 느려지게되는 단점이 있다. 기존 알고리즘의 이러한 단점을 극복하기 위해 기존의 GP 알고리즘에 Conjugate Gradient 법을 결합시켜 보다 효율적인 경로기반 통행배정 알고리즘을 구축하였다. 이는 최적해 근처에서 더욱 정확한 이동방향을 결정하여 빠른 시간 내에 최적해를 도출해 내도록 하기 위한 것이다. 또한, 구축된 알고리즘을 가로망에 적용, 그 효율성을 검증하여 Conjugate Gradient 법이 통행 배정 모형의 사용자 평형 모형에서와 같은 목적함수의 경우에서도 매우 빠른 수렴을 위해 유용하게 쓰일 수 있다는 것을 보였다.

• 전산구조공학
• /
• 제8권4호
• /
• pp.99-106
• /
• 1995

구조물의 최적설계에 대한 일반적인 개념이 제시되며 Gradient Projection법과 설계민감도 해석을 사용한 최적화 알고리즘이 논의된다. 6층 평면 철골 구조물의 최소무게설계의 예를 통해 이 알고리즘의 적용을 예시한다. 시스템 전체의 비용분석뿐만 아니라 임계 비용 및 설계민감도 해석과 같은 이 알고리즘의 장점에 관하여 설명한다.