• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제16권4호
• /
• 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.

• 한국통계학회:학술대회논문집
• /
• 한국통계학회 2003년도 춘계 학술발표회 논문집
• /
• pp.85-90
• /
• 2003

In this paper we develop a method for finding optimal ordering of K statistical models. This is based on a dependent paired comparison experimental arrangement whose results can naturally be represented by a completely oriented graph (also so called tournament graph). Introducing preference probabilities, strong transitivity conditions, and an optimal criterion to the graph, we show that a Hamiltonian path obtained from row sum ranking is the optimal ordering. Necessary theories involved in the method and computation are provided. As an application of the method, generalized variances of K multivariate normal populations are compared by a Bayesian approach.

• 경영과학
• /
• 제30권1호
• /
• pp.167-181
• /
• 2013

We investigate the dynamic asset allocation problem under inflation risk when the wealth of an investor is constrained with minimum requirements. To capture the investor's risk preference, the CRRA utility function is considered and he maximizes his expected utility at predetermined date of the refund by participation in the financial market. The financial market is supposed to consist of three kinds of financial instruments which are a risk free asset, a risky asset, and an index bond. The role of an index bond is managing inflation risk represented by price process. The optimal wealth and the optimal asset allocation are derived explicitly by using the method to get the European call option pricing formula. From the numerical results, it is confirmed that the investments on index bond is high when the investor's wealth level is low. However, as his wealth increases, the investments on index bond decreases and he invests on risky asset more. Furthermore, the minimum wealth constraint induces lower investment on risky asset but the effect of the constraints is reduced as the wealth level increases.

• 산업경영시스템학회지
• /
• 제33권2호
• /
• pp.1-8
• /
• 2010

A vendor selection problem consists of two different kinds of decision making. First one is to choose the best suppliers among all possible suppliers and the next is to allocate the optimal quantities of orders among the selected vendors. In this study, an integration of the technique for order preference by similarity to ideal solution (TOPSIS) and a multi-objective mixed integer programming (MOMIP) is developed to account for all qualitative and quantitative factors which are used to evaluate and choose the best group of vendors and to decide the optimal order quantity for each vendor. A solution methodology for the vendor selection model of multiple-vendor, multiple-item with multiple decision criteria and in respect to finite vendor capacity is presented.

• 산업경영시스템학회지
• /
• 제22권49호
• /
• pp.11-21
• /
• 1999

The purpose of this study is to present the method to select optimal size for the industrial products which are closely related to human's body size. For this purpose, human factors such as body characteristics, body features, and preference in product selection which needs to be considered in setting standards were analyzed. This analysis is to select optimal size to minimize losses caused by the difference of size between demand by the customers and supply from the manufacturers. Using loss function, repetitive calculation process algorithm by using bisearch method was applied in selecting the sizes of demand and supply which minimize the total expected losses. For cumulative normal distribution probability, IMSL routine DNORDF was used. In case study, comparison has been made between the result which was calculated using presented algorithm and the results calculated by the process currently used by KS and ISO by measuring aged women's body size in human factors side and sorting them through the factor analysis and cluster analysis for feature factor extraction. Thus, they can be used as a basis for establishing industrial product standards.

• 한국실내디자인학회논문집
• /
• 제40호
• /
• pp.86-93
• /
• 2003

Korean traditionally prefer a house with a southern exposure to other styles. Moreover, pretty part of dwelling style in Korea is apartment building. But generally, there are much more north oriented rooms than opposite one in apartment. Generally, most Korean tends to consider the direction of the room as an important factor to decide the arrangement of their furniture. Also, the location of window, door, bed, and closet can affect on the decision-making process of furniture layout. So, the closet position should be determined by not only window direction and door position, but also the furniture arrangement preference of residents. After considering the window and door position, there are 16 alternatives in a room. (Multiplied 4 window directions by 4 door positions under the assumption that the window and the door cannot be on the same side of the wall.) Also, this study analyzed each room style ratio considering the symmetry of each plan by comparing the apartment building plan. This paper studied the bedroom style by analyzing the 180 apartment building plans and survey of furniture layout; the furniture in this study represents bed and desk those are mainly laid in Korean bedroom. Finally, this paper suggested the methodology to find optimal alternatives of furniture and closet layout for high-rise apartment residents, based on the door and window position, and the furniture layout preference of residents.

• 대한산업공학회지
• /
• 제38권1호
• /
• pp.57-66
• /
• 2012

The paper deals with the development of a model for group decision making under multiple criteria. The Multi-criteria group decision making (MCGDM) is the process to determine the best compromise solution in a set of competing alternatives that are evaluated by decision makers having their own preferences on conflicting objectives. For MCGDM, we propose a Mixed-Integer Programming (MIP) model that implements a revised median approach by noticing that the original median approach cannot consider the willingness to reject and the indifferent preference conditions. The proposed MIP model tries to select a common best Pareto-optimal solution by maximizing the overall desirability considering the willingness to reject and the indifferent preference that represent the tolerance measure of each decision maker. To evaluate the effectiveness of the proposed model, we compared the results of the proposed model with those of the median approach. The results showed that the proposed MIP model produces more realistic and better compromised alternative by incorporating the decision maker's willingness to reject and the indifferent preferences over each criteria.

• 산업공학
• /
• 제11권3호
• /
• pp.143-153
• /
• 1998

Warning control is one of the most important components in driver's cabin for achieving safe driving. The purpose of this study is to suggest ergonomically optimal location and type of passenger car's warning control. An experiment was conducted using driving simulator, in which nine locations and two types of warning controls -push button, rocker switch- were employed as experimental variables, and response time for warning signal and preference for locations and types of warning controls were measured as dependent variables. The results showed that response time for warning signal was the shortest when warning control was located at the middle left corner of the center fascia, and was the second at the middle center of the center fascia. Preference for warning signal was the highest at the middle left comer of the center fascia, and was the second between the steering wheel column and the center fascia. Although push button was not preferred to rocker switch, response time was shorter for push button than for rocker switch. It was suggested from these results that warning control with the type of push button should he located at among the middle left corner of the center fascia, the middle center of the center fascia, and between the steering wheel column and the center fascia.

• 한국식생활문화학회지
• /
• 제12권3호
• /
• pp.265-274
• /
• 1997

This study was conducted in ordor to find the most suitable conditions for producing the Doenjang with optimal odor compound contents. Three sample groups with the different preparing methods -Doenjang that has not gone through the soy sauce separation process (Doenjang A), Doenjang that has gone through the soy sauce separation process; Meju-20% salt water ratio of 1 : 4 (Doenjang B), and that with the ratio of 1.3 : 4 (Doenjang C)- were tested during different ripening periods. Odor compound contents were analyzed through Solvent Extraction Method and Simultaneous Steam Distillation Extraction (SDE). The number of odor compounds was greatest in Doenjang A and during the mid to late stage in each groups. In the sensory evaluation of Doenjang odors, Doenjang A received the highest scores in the categories of overall preferences, while Doenjang C got the lowest scores. Individual odor didn't vary significantly during ripening periods, but the overall odor and taste preference was highest in the samples ripened for 75 to 120 days. Stepwise multiple regression analysis of major odor compounds of Doenjang revealed that benzeneacetaldehyde is the major explanatory variable for offensive odor. Benzeneethanol, 3-methylthio-propanal and 4-methyl-phenol are the explanatory variables for salty odor, nutty odor and rancid odor, respectively. Odor compounds that contribute to the overall odor preference varied from the compounds that affect the taste preference.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권11호
• /
• pp.5234-5251
• /
• 2018

It is well-known that virtualization technology can bring many benefits not only to users but also to service providers. From the view of system security and resource utility, higher resource sharing degree and higher system reliability can be obtained by the introduction of virtualization technology in distributed cloud. The small size time-sharing multiplexing technology which is based on virtual machine in distributed cloud platform can enhance the resource utilization effectively by server consolidation. In this paper, the concept of memory block and user satisfaction is redefined combined with user requirements. According to the unbalanced memory resource states and user preference requirements in multi-virtual machine environments, a model of proper memory resource allocation is proposed combined with memory block and user satisfaction, and at the same time a memory optimization allocation algorithm is proposed which is based on virtual memory block, makespan and user satisfaction under the premise of an orderly physical nodes states also. In the algorithm, a memory optimal problem can be transformed into a resource workload balance problem. All the virtual machine tasks are simulated in Cloudsim platform. And the experimental results show that the problem of virtual machine memory resource allocation can be solved flexibly and efficiently.