• Journal of the Korea Society for Simulation
• /
• v.19 no.2
• /
• pp.91-98
• /
• 2010

Conventional path-finding has focused on finding short collision-free paths. However, as computer games become more sophisticated, it is required to take tactical information like ambush points or lines of enemy sight into account. One way to make this information have an effect on path-finding is to represent a heuristic function of a search algorithm as a weighted sum of tactics. In this paper we consider the problem of learning heuristic to optimize path-finding based on given tactical information. What is meant by learning is to produce a good weight vector for a heuristic function. Training examples for learning are given by a game level-designer and will be compared with search results in every search level to update weights. This paper proposes a learning algorithm integrated with search for tactical path-finding. The perceptron-like method for updating weights is described and a simulation tool for implementing these is presented. A level-designer can mark desired paths according to characters' properties in the heuristic learning tool and then it uses them as training examples to learn weights and shows traces of paths changing along with weight learning.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.16 no.4
• /
• pp.270-275
• /
• 2016

We present an approach for acceleration actor-critic algorithm for reinforcement learning with continuous action space. Actor-critic algorithm has already proved its robustness to the infinitely large action spaces in various high dimensional environments. Despite that success, the main problem of the actor-critic algorithm remains the same-speed of convergence to the optimal policy. In high dimensional state and action space, a searching for the correct action in each state takes enormously long time. Therefore, in this paper we suggest a search accelerating function that allows to leverage speed of algorithm convergence and reach optimal policy faster. In our method, we assume that actions may have their own distribution of preference, that independent on the state. Since in the beginning of learning agent act randomly in the environment, it would be more efficient if actions were taken according to the some heuristic function. We demonstrate that heuristically-accelerated actor-critic algorithm learns optimal policy faster, using Educational Process Mining dataset with records of students' course learning process and their grades.

• Industrial Engineering and Management Systems
• /
• v.6 no.2
• /
• pp.136-145
• /
• 2007

The selection of suppliers and the determination of order quantities to be placed with those suppliers are important decisions in a supply chain. In this research, a non-linear mixed integer programming model is presented to select suppliers and determine the order quantities. The model considers the purchasing cost which takes into account quantity discount, the cost of transportation, the fixed cost for establishing suppliers, the cost for holding inventory, and the cost of receiving poor quality parts. The capacity constraints for suppliers, quality and lead-time requirements for the parts are also taken into account in the model. Since the purchasing cost, which is a decreasing step function of order quantities, introduces discontinuities to the non-linear objective function, it is not easy to employ traditional optimization methods. Thus, a heuristic algorithm, called particle swarm optimization (PSO), is used to find the (near) optimal solution. However, PSO usually generates initial solutions randomly. To improve the PSO solution quality, a heuristic procedure is proposed to find an initial solution based on the average unit cost including transportation, purchasing, inventory, and poor quality part cost. The results show that PSO with the proposed initial solution heuristic provides better solutions than those with PSO algorithm only.

• Journal of Intelligence and Information Systems
• /
• v.15 no.3
• /
• pp.137-150
• /
• 2009

This paper proposes the hybrid heuristic method to apply the opportunity time to solve the vehicle routing and scheduling problem with time constraints(VRSPTW). The opportunity time indicates the idle time which remains after the vehicle performs the unloading service required by each customer's node. In this proposed heuristic, we add the constraints to VRSPTW model for the opportunity time. We also obtain the initial solution by applying the cost evaluation function to the insertion strategy considering the opportunity time. In addition, we improve the former result by applying the opportunity time to the tabu search strategy by swapping the customer's node. Finally, we suggest the construction strategies of initial routing which can efficiently acquire the nearest optimal solution from various types of data in terms of geographical condition, scheduling horizon and vehicle capacity. Our experiment show that our heuristic can get the nearest optimal solution more efficiently than the Solomon's I1 heuristic.

• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.1
• /
• pp.49-62
• /
• 2024

In this paper, a sustainable closed-loop supply chain (SCLSC) network model is proposed for domestic mobile phone industry. Economic, environmental and social factors are respectively considered for reinforcing the sustainability of the SCLSC network model. These three factors aim at minimizing total cost, minimizing total amount of CO₂ emission, and maximizing total social influence resulting from the establishment and operation of facilities at each stage of the SCLSC network model. Since they are used as each objective function in modeling, the SCLSC network model can be a multi-objective optimization problem. A mathematical formulation is used for representing the SCLSC network model and a hybrid meta-heuristic approach is proposed for efficiently solving it. In numerical experiment, the performance of the proposed hybrid meta-heuristic approach is compared with those of conventional meta-heuristic approaches using some scales of the SCLSC network model. Experimental results shows that the proposed hybrid meta-heuristic approach outperforms conventional meta-heuristic approaches.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.14 no.3
• /
• pp.1-8
• /
• 2015

In transportation networks, searching speed and result accuracy are becoming more critical on searching minimum path algorithm. Current $A^*$ algorithm has a big advantage of high searching speed. However, it has disadvantage of complicated searching network and low accuracy rate of finding the minimum path algorithm. Therefore, this study developed $A^*$ algorithm's heuristic function and focused on improving it's disadvantages. Newly developed function in this study contains the area concept, not the line concept. During the progress, this study adopts the idea of a heavier node that remains lighter to the target node is better that the lighter node that becomes heavier when it is connected to the other. Lastly, newly developed algorithm has the feedback function, which allows the larger accuracy value of heuristic than before. This developed algorithm tested on real network, and proved that developed algorithm is useful.

• Communications of the Korean Mathematical Society
• /
• v.9 no.2
• /
• pp.327-329
• /
• 1994

The Dirichlet problem (DP) on the upper half plane {z = x ＋ iy : y > 0} is to find a real-valued harmonic function u(x, y) satisfying u(x, 0) = g(x) almost everywhere for some reasonably nice function g defined on the real line, which is called the data on the boundary for (DP).(omitted)

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1993.06a
• /
• pp.1151-1154
• /
• 1993

The purpose of the paper is to explain some heuristic, common sense suppositions of fuzzy control. It is shown that Fuzzy Control is a kind of quasilinear interpolation of prototypes. Control function can be sufficiently exact represented as piecewise-linear function. The best interpolation is connected with normalized intersected fuzzy sets.

• Proceedings of the KIEE Conference
• /
• 1991.07a
• /
• pp.34-36
• /
• 1991

The idea of expert control is to incoporate a rule based expert system in a feedback control system. In this paper, we present some heuristic rules about input requlation and supervision and turning for D.C servo motor speed control in variable. The expert auto-turning PID controller which heuristic rules are used as an element of the feedback control system is implemented with the numerical algorithms and heuristic logics. The accurate control function is confirmed by computer simulation.

• Korean Management Science Review
• /
• v.17 no.1
• /
• pp.41-54
• /
• 2000

in the real world situations that some jobs need be processed only on certain limited machines frequently occur due to the capacity restrictions of machines such as tools fixtures or material handling equipment. In this paper we consider n-job non-preemptive and m parallel machines scheduling problem having two machines group. The objective function is to minimize the sum of earliness and tardiness with different release times and due dates. The problem is formulated as a mixed integer programming problem. The problem is proved to be Np-complete. Thus a heuristic is developed to solve this problem. To illustrate its suitability and efficiency a proposed heuristic is compared with a genetic algorithm and tabu search for a large number of randomly generated test problems in ship engine assembly shop. Through the experimental results it is showed that the proposed algorithm yields good solutions efficiently.