• 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.

• 한국전산구조공학회논문집
• /
• 제37권3호
• /
• pp.187-195
• /
• 2024

본 논문에서는 평활화 유한요소법(Smoothed finite element method)을 도입한 응력 기반 구배 탄성론(Gradient elasticity)의 2차원 경계치 문제에 대한 연구를 수행하였다. 구배 탄성론은 기존 탄성론에서는 표현할 수 없는 미소규모의 크기 의존적인 기계적 거동을 설명하기 위해 제안되었다. 구배 탄성체론에서 고차 미분 방정식을 두 개의 2차 미분 방정식으로 분할하는 Ru-Aifantis 이론을 사용하기 때문에 평활화 유한요소법에 적용이 가능하게 된다. 본 연구에서 경계치 문제를 해결하기 위해 평활화 유한 요소 프레임워크에 스태거드 방식(Staggered scheme)을 사용하여 국부 변위장과 비국부 응력장을 평활화 영역 및 요소에서 각각 계산하였다. 구배 탄성에서 중요한 변수인 내부 길이 척도의 영향을 측정하기 위해 일련의 수치 예제를 수행하였다. 수치 해석 결과는 제안한 기법이 내부 길이 척도에 따라 균열 선단과 전위 선에 나타나는 응력 집중을 완화할 수 있음을 보여준다.

• 디지털산업정보학회논문지
• /
• 제5권4호
• /
• pp.117-126
• /
• 2009

The recent interest in sensor networks has led to a number of routing schemes that use the limited resources available at sensor nodes more efficiently. These schemes typically try to find the minimum energy path to optimize energy usage at a node. Some schemes, however, are prone to unbalance of the traffic and energy. To solve this problem, we propose a novel solution: a gradient-field approach which takes account of the minimum cost data delivery, energy consumption balancing, and traffic equalization. We also modify the backoff-based cost field setup algorithm to establish our gradient-field based sensor network and give the algorithm. Simulation results show that the overhead of routing establishment obtained by our algorithm is much less than the one obtained by Flooding. What's more, our approach guarantees the basic Quality of Service (QoS) without extra spending.

• 로봇학회논문지
• /
• 제17권1호
• /
• pp.58-67
• /
• 2022

In this paper, we propose an approach resolving inaccuracy of the low-cost redundant manipulator workspace with low encoder and low stiffness. When the manipulators are manufactured with low-cost encoders and low-cost links, the robots can run into workspace inaccuracy issues. Furthermore, trajectory generation based on conventional forward/inverse kinematics without taking into account inaccuracy issues will introduce the risk of end-effector fluctuations. Hence, we propose an optimization for the trajectory generation method based on the DDPG (Deep Deterministic Policy Gradient) algorithm for the low-cost redundant manipulators reaching the target position in Euclidean space. We designed the DDPG algorithm minimizing the distance along with the jacobian condition number. The training environment is selected with an error rate of randomly generated joint spaces in a simulator that implemented real-world physics, the test environment is a real robotic experiment and demonstrated our approach.

• 소성∙가공
• /
• 제15권8호
• /
• pp.544-549
• /
• 2006

In this work, we have employed the strain gradient plasticity theory to investigate the effect of material size on the deformation behavior in metal forming process. Flow stress is expressed in terms of strain, strain gradient (spatial derivative of strain) and intrinsic material length. The least square method coupled with strain gradient plasticity was used to calculate the components of strain gradient at each element of material. For demonstrating the size effect, the proposed approach has been applied to plane compression process and micro rolling process. Results show when the characteristic length of the material comes to the intrinsic material length, the effect of strain gradient is noteworthy. For the microcompression, the additional work hardening at higher strain gradient regions results in uniform distribution of strain. In the case of micro-rolling, the strain gradient is remarkable at the exit section where the actual reduction of the rolling finishes and subsequently strong work hardening take places at the section. This results in a considerable increase in rolling force. Rolling force with the strain gradient plasticity considered in analysis increases by 20% compared to that with conventional plasticity theory.

• Steel and Composite Structures
• /
• 제46권2호
• /
• pp.153-173
• /
• 2023

Rectangular concrete-filled steel tubular (RCFST) column, a type of concrete-filled steel tubular (CFST), is widely used in compression members of structures because of its advantages. This paper proposes a robust machine learning-based framework for predicting the ultimate compressive strength of RCFST columns under both concentric and eccentric loading. The gradient boosting neural network (GBNN), an efficient and up-to-date ML algorithm, is utilized for developing a predictive model in the proposed framework. A total of 890 experimental data of RCFST columns, which is categorized into two datasets of concentric and eccentric compression, is carefully collected to serve as training and testing purposes. The accuracy of the proposed model is demonstrated by comparing its performance with seven state-of-the-art machine learning methods including decision tree (DT), random forest (RF), support vector machines (SVM), deep learning (DL), adaptive boosting (AdaBoost), extreme gradient boosting (XGBoost), and categorical gradient boosting (CatBoost). Four available design codes, including the European (EC4), American concrete institute (ACI), American institute of steel construction (AISC), and Australian/New Zealand (AS/NZS) are refereed in another comparison. The results demonstrate that the proposed GBNN method is a robust and powerful approach to obtain the ultimate strength of RCFST columns.

• 정보처리학회논문지B
• /
• 제15B권1호
• /
• pp.25-36
• /
• 2008

본 논문은 조명의 변화가 심한 영상에서 손 형태를 안정적으로 인지하는 기법에 관한 것이다. 제안한 방법은 HSI 색상공간에서 색상(Hue) 및 색상 기울기(Hue-Gradient)를 기반으로 정의된 배경모델을 구축하고, 실시간으로 입력되는 영상과의 배경차분(background subtraction)기법을 이용하여 배경과 손을 구분한다. 추출된 손의 영역으로부터 18가지의 특징요소를 추출하고 이를 기반으로 다중클래스 SVM(Support Vector Machine) 학습 기법을 사용하여 손의 형태를 인지한다. 제안 기법은 색상 기울기를 배경 차분에 적용함으로써, 조명 환경이 배경 모델의 조명과 다르게 급격한 변화가 이루어졌을 때도 안정적으로 손의 윤곽정보를 추출할 수 있도록 하였다. 또한, 실시간 처리를 저해하는 복잡한 손의 특성정보 대신, OBB의 크기에 대하여 정규화된 두 개의 고유값과 객체 기반 바운딩 박스(OBB)를 구성하는 16개 세부 영역에서의 손 윤곽픽셀의 개수를 손의 특성정보로 사용하였다. 본 논문에서는 급격한 조명 변화 상황에서 기존 RGB 색상요소를 기반으로 하는 배경차분법과 색상을 기반으로 하는 배경차분법, 본 논문에서 제안하는 색상 기울기 기반 배경 차분법의 결과를 비교함으로써 제안 기법의 안정성을 입증하였다. 6명의 실험대상자의 1부터 9까지의 수지화 2700개의 영상으로부터 손 특성 정보를 추출하고 이에 대하여 훈련을 통한 학습 모델을 생성하였다. 학습모델을 기반으로 실험자 6인의 손 형태 1620개의 데이터에 대하여 인지 실험을 실시하여 92.6%에 이르는 손 형태 인식 성공률을 얻었다.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제24권1호
• /
• pp.1-22
• /
• 2020

An accurate and efficient gradient estimation method on unstructured grid is presented by proposing a switching process between two Least-Square methods. Diverse test cases show that the gradient estimation by Least-Square methods exhibit better characteristics compared to Green-Gauss approach. Based on the investigation, switching between the two Least-Square methods, whose merit complements each other, is pursued. The condition number of the Least-Square matrix is adopted as the switching criterion, because it shows clear correlation with the gradient error, and it can be easily calculated from the geometric information of the grid. To illustrate switching process on general grid, condition number is analyzed using stencil vectors and trigonometric relations. Then, the threshold of switching criterion is established. Finally, the capability of Switching Weighted Least-Square method is demonstrated through various two- and three-dimensional applications.

• 센서학회지
• /
• 제27권4호
• /
• pp.209-215
• /
• 2018

Recent research in animal behavior has shown that gradient information plays an important role in finding food and home. It is also important in optimization of performance because it indicates how the inputs should be adjusted for maximization/minimization of a performance index. We introduce perturbation as an additional input to obtain gradient information. Unlike the typical approach of calculating the gradient from the derivative, the proposed processing is very robust to noise since it is performed as a summation. Experimental results prove the validity of the process of spatial gradient acquisition. Quantitative indices for measuring the effect of the amplitude and the frequency are developed based on linear regression analysis. Drones are very useful for environmental monitoring and an autonomous path planning is required for unstructured environment. Guiding the drone for finding the origin of the interested physical property is done by estimating the gradient of the sensed value and generating the drone trajectories in the direction which maximizes the sensed value. Simulation results show that the proposed method can be successfully applied to identify the source of the physical quantity of interest by utilizing it for path planning of an autonomous drone in 3D environment.

• Structural Monitoring and Maintenance
• /
• 제7권2호
• /
• pp.69-84
• /
• 2020

Based on differential quadrature method (DQM) and nonlocal strain gradient theory (NSGT), forced vibrations of a porous functionally graded (FG) scale-dependent beam in thermal environments have been investigated in this study. The nanobeam is assumed to be in contact with a moving point load. NSGT contains nonlocal stress field impacts together with the microstructure-dependent strains gradient impacts. The nano-size beam is constructed by functionally graded materials (FGMs) containing even and un-even pore dispersions within the material texture. The gradual material characteristics based upon pore effects have been characterized using refined power-law functions. Dynamical deflections of the nano-size beam have been calculated using DQM and Laplace transform technique. The prominence of temperature rise, nonlocal factor, strain gradient factor, travelling load speed, pore factor/distribution and elastic substrate on forced vibrational behaviors of nano-size beams have been explored.