• Journal of Korea Multimedia Society
• /
• v.3 no.4
• /
• pp.356-364
• /
• 2000

Block based fast motion estimation algorithm use the fixed search pattern to reduce the search point, and are based on the assumption that the error in the mean absolute error space monotonically decreases to the global minimum. Therefore, in case of many local minima in a search region we are likely to find local minima instead of the global minimum and highly rely on the initial search points. This situation is evident in the motion boundary. In this paper we define the candidate regions within the search region using the motion information of the neighbor blocks and we propose the multiple local search method (MLSM) which search for the solution throughout the candidate regions to reduce the possibilities of isolation to the local minima. In the MLSM we mark the candidate region in the search point map and we avoid to search the candidate regions already visited to reduce the calculation. In the simulation results the proposed method shows more excellent results than that of other gradient based method especially in the search of motion boundary. Especially, in PSNR the proposed method obtains similar estimate accuracy with the significant reduction of search points to that of full search.

• Proceedings of the Computational Structural Engineering Institute Conference
• /
• 2005.04a
• /
• pp.351-358
• /
• 2005

Many methods have been developed and are in use for structural size optimization problems, In which the cross-sectional areas or sizing variables are usually assumed to be continuous. In most practical structural engineering design problems, however, the design variables are discrete. This paper proposes an efficient optimization method for structures with discrete-sized variables based on the harmony search (HS) meta-heuristic algorithm. The recently developed HS algorithm was conceptualized using the musical process of searching for a perfect state of harmony. It uses a stochastic random search instead of a gradient search so that derivative information is unnecessary In this paper, a discrete search strategy using the HS algorithm is presented in detail and its effectiveness and robustness, as compared to current discrete optimization methods, are demonstrated through a standard truss example. The numerical results reveal that the proposed method is a powerful search and design optimization tool for structures with discrete-sized members, and may yield better solutions than those obtained using current method.

• Journal of Korea Water Resources Association
• /
• v.56 no.1
• /
• pp.63-74
• /
• 2023

High-reliability prediction of dam inflow is necessary for efficient dam operation. Recently, studies were conducted to predict the inflow of dams using Multi Layer Perceptron (MLP). Existing studies used the Gradient Descent (GD)-based optimizer as the optimizer among MLP operators to find the optimal correlation between data. However, the GD-based optimizers have disadvantages in that the prediction performance is deteriorated due to the possibility of convergence to the local optimal value and the absence of storage space. This study improved the shortcomings of the GD-based optimizer by developing Adaptive moments combined with Improved Harmony Search (AdamIHS), which combines Adaptive moments among GD-based optimizers and Improved Harmony Search (IHS). In order to evaluate the learning and prediction performance of MLP using AdamIHS, Daecheong Dam inflow was learned and predicted and compared with the learning and prediction performance of MLP using GD-based optimizer. Comparing the learning results, the Mean Squared Error (MSE) of MLP, which is 5 hidden layers using AdamIHS, was the lowest at 11,577. Comparing the prediction results, the average MSE of MLP, which is one hidden layer using AdamIHS, was the lowest at 413,262. Using AdamIHS developed in this study, it will be possible to show improved prediction performance in various fields.

• Proceedings of the KIEE Conference
• /
• 1998.07b
• /
• 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.

• Advances in Energy Research
• /
• v.6 no.2
• /
• pp.91-101
• /
• 2019

This paper proposes an efficient data-driven approach to build models for predicting energy consumption in buildings. Data used in this research is collected by installing humidity and temperature sensors at different locations in a building. In addition to this, weather data from nearby weather station is also included in the dataset to study the impact of weather conditions on energy consumption. One of the main emphasize of this research is to make feature selection independent of domain knowledge. Therefore, to extract useful features from data, two different approaches are tested: one is feature selection through principal component analysis and second is relative importance-based feature selection in original domain. The regression model used in this research is gradient boosting regression and its optimal parameters are chosen through a two staged coarse-fine search approach. In order to evaluate the performance of model, different performance evaluation metrics like r2-score and root mean squared error are used. Results have shown that best performance is achieved, when relative importance-based feature selection is used with gradient boosting regressor. Results of proposed technique has also outperformed the results of support vector machines and neural network-based approaches tested on the same dataset.

• Structural Engineering and Mechanics
• /
• v.88 no.4
• /
• pp.301-309
• /
• 2023

In this paper, a new modified conjugate gradient (MCG) method is presented which is based on a new gradient regularizer, and this method is used to identify the dynamic load on airfoil structure without and with considering random structure parameters. First of all, the newly proposed algorithm is proved to be efficient and convergent through the rigorous mathematics theory and the numerical results of determinate dynamic load identification. Secondly, using the perturbation method, we transform uncertain inverse problem about force reconstruction into determinate load identification problem. Lastly, the statistical characteristics of identified load are evaluated by statistical methods. Especially, this newly proposed approach has successfully solved determinate and uncertain inverse problems about dynamic load identification. Numerical simulations validate that the newly developed method in this paper is feasible and stable in solving load identification problems without and with considering random structure parameters. Additionally, it also shows that most of the observation error of the proposed algorithm in solving dynamic load identification of deterministic and random structure is respectively within 11.13%, 20%.

• 한국전산유체공학회:학술대회논문집
• /
• 2005.10a
• /
• pp.163-167
• /
• 2005

A research to evaluate efficiency of design optimization was performed for aerodynamic design optimization problem in distributed computing environment. The aerodynamic analyses which take most of computational work during design optimization were divided into several jobs and allocated to associated PC clients through network. This is not a parallel process based on domain decomposition rather than a simultaneous distributed-analyses process using network-distributed computers. GBOM(gradient-based optimization method), SAO(Sequential Approximate Optimization) and RSM(Response Surface Method) were implemented to perform design optimization of transonic airfoil and to evaluate their efficiencies. One dimensional minimization followed by direction search involved in the GBOM was found an obstacle against improving efficiency of the design process in distributed computing environment. The SAO was found quite suitable for the distributed computing environment even it has a handicap of local search. The RSM is apparently the fittest for distributed computing environment, but additional trial and error works needed to enhance the reliability of the approximation model are annoying and time-consuming so that they often impair the automatic capability of design optimization and also deteriorate efficiency from the practical point of view.

• Journal of computational fluids engineering
• /
• v.11 no.2 s.33
• /
• pp.19-24
• /
• 2006

A research to evaluate the efficiency of design optimization was carried out for aerodynamic design optimization problem in distributed computing environment. The aerodynamic analyses which take most of computational work during design optimization were divided into several jobs and allocated to associated PC clients through network. This is not a parallel process based on domain decomposition in a single analysis rather than a simultaneous distributed-analyses using network-distributed computers. GBOM(gradient-based optimization method), SAO(Sequential Approximate Optimization) and RSM(Response Surface Method) were implemented to perform design optimization of transonic airfoils and evaluate their efficiencies. dimensional minimization followed by direction search involved in the GBOM was found an obstacle against improving efficiency of the design process in the present distributed computing system. The SAO was found fairly suitable for the distributed computing environment even it has a handicap of local search. The RSM is apparently the most efficient algorithm in the present distributed computing environment, but additional trial and error works needed to enhance the reliability of the approximation model deteriorate its efficiency from the practical point of view.

• Korean Journal of Air-Conditioning and Refrigeration Engineering
• /
• v.7 no.4
• /
• pp.566-576
• /
• 1995

A Computational code has been developed in order to design axial fans by the numerical optimization techniques incorporated with flow analysis code solving three-dimensional Navier-Stokes equation. The steepest descent method and the conjugate gradient method are used to look for the search direction in the design space, and the golden section method is used for one-dimensional search. To solve the constrained optimization problem, sequential unconstrained minimization technique, SUMT, is used with imposed quadratic extended interior penalty functions. In the optimization of two-dimensional cascade design, the ratio of drag coefficient to lift coefficient is minimized by the design variables such as maximum thickness, maximum ordinate of camber and chord wise position of maximum ordinate. In the application of this numerical optimization technique to the design of an axial fan, the efficiency is maximized by the design variables related to the sweep angle distributed by quadratic function along the hub to tip of fan.

• Journal of information and communication convergence engineering
• /
• v.13 no.2
• /
• pp.123-131
• /
• 2015

Training neural networks is a complex task with great importance in the field of supervised learning. In the training process, a set of input-output patterns is repeated to an artificial neural network (ANN). From those patterns weights of all the interconnections between neurons are adjusted until the specified input yields the desired output. In this paper, a new hybrid algorithm is proposed for global optimization of connection weights in an ANN. Dynamic swarms are shown to converge rapidly during the initial stages of a global search, but around the global optimum, the search process becomes very slow. In contrast, the gradient descent method can achieve faster convergence speed around the global optimum, and at the same time, the convergence accuracy can be relatively high. Therefore, the proposed hybrid algorithm combines the dynamic particle swarm optimization (DPSO) algorithm with the backpropagation (BP) algorithm, also referred to as the DPSO-BP algorithm, to train the weights of an ANN. In this paper, we intend to show the superiority (time performance and quality of solution) of the proposed hybrid algorithm (DPSO-BP) over other more standard algorithms in neural network training. The algorithms are compared using two different datasets, and the results are simulated.