• Korean Management Science Review
• /
• v.32 no.1
• /
• pp.83-100
• /
• 2015

We analyze a duopoly competition when two firms face input cost increases. The objective of this study is to determine the firms' optimal strategy between a price increase and downsizing under conditions of a spatially differentiated market and consumers' diminishing utility on the product size. We develop a theoretical model of two competing firms offering homogenous products using the standard Hotelling model to determine how firms' optimal strategies change when facing input cost increases. In this paper, there are two types of duopoly competitions: symmetric and asymmetric. In the symmetric case, the two firms have the same marginal cost and are producing and selling identical products. In the asymmetric case, the two firms have different marginal costs. The results show that the optimal strategy decision depends on the size of the input cost increase and the cost differences between the two firms. We find that when two firms are asymmetric (i.e., they have different marginal costs), the two firms might choose asymmetric pairs of strategies in equilibrium under certain conditions. When the cost differences between the two firms are sufficiently large and the cost increase is sufficiently small, the cost leader chooses price increase, and the cost-disadvantaged firm chooses downsizing in equilibrium. This asymmetric strategy reduces price competition between two firms, and consumers are better off. When the cost differences between the two firms are sufficiently large, downsizing is the dominant strategy for the cost-disadvantaged firm. The cost-disadvantaged firm finds it more profitable to reduce the product size than to increase its price to reduce price competition, because consumers prefer downsizing to price increases. This paper might be a good starting point for further analytical research in this area.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2008.04a
• /
• pp.442-448
• /
• 2008

This paper presents an approach to define an optimal piezoactuator length to actively control structural vibration. The optimal ratio of the piezoactuator length against beam length when a pair of piezoceramic actuator and accelerometer is used to suppress unwanted vibration with direct velocity feedback (DVFB) control strategy is not clearly defined so far. It is well known that direct velocity feedback (DVFB) control can be very useful when a pair of sensor and actuator is collocated on structures with a high gain and excellent stability. It is considered that three different collocated pairs of piezoelectric actuators (20, 50 and 100 mm) and accelerometers installed on three identical clamped-clamped beams (300 * 20 * 1 mm). The response of each sensor-actuator pair requires strictly positive real (SPR) property to apply a high feedback gain. However the length of the piezoactuator affects SPR property of the sensor-actuator response. Intensive simulation and experiment shows the effect of the actuator length variation is strongly related with the frequency range of SPR property. A shorter actuator gave a wider SPR frequency range as a longer one had a narrower range. The shorter actuator showed limited control performance in spite of a higher gain was applied because the actuation force was relatively small. Thus an optimal length ratio (actuator length/beam length) was suggested to obtain relevant performance with good stability with DVFB strategy. The result of this investigation could give important information in the design of active control system to suppress unwanted vibration of smart structures with piezoelectric actuators and accelerometers.

• Journal of Communications and Networks
• /
• v.17 no.2
• /
• pp.184-197
• /
• 2015

This paper builds on a recent method, chain routing with even energy consumption (CREEC), for designing a wireless sensor network with chain topology and for scheduling the communication to ensure even average energy consumption in the network. In here a new suboptimal design is proposed and compared with the CREEC design. The chain topology in CREEC is reconfigured after each group of n converge-casts with the goal of making the energy consumption along the new paths between the nodes in the chain as even as possible. The new method described in this paper designs a single near-optimal Hamiltonian circuit, used to obtain multiple chains having only the terminal nodes different at different converge-casts. The advantage of the new scheme is that for the whole life of the network most of the communication takes place between same pairs of nodes, therefore keeping topology reconfigurations at a minimum. The optimal scheduling of the communication between the network and base station in order to maximize network lifetime, given the chosen minimum length circuit, becomes a simple linear programming problem which needs to be solved only once, at the initialization stage. The maximum lifetime obtained when using any combination of chains is shown to be upper bounded by the solution of a suitable linear programming problem. The upper bounds show that the proposed method provides near-optimal solutions for several wireless sensor network parameter sets.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.34 no.6
• /
• pp.1765-1777
• /
• 2014

This study evaluated PMM (Probability Matching Method) for parameter estimation of the Z - R relation. As a first step, the sensitivity analysis was done to decide the threshold number of data pairs and the data interval for the development of a histogram. As a result, it was found that at least 1,000 number of data pairs are required to apply the PMM for the parameter estimation. This amount of data is similar to that collected for two hours. Also, the number of intervals for the histogram was found to be at least 100. Additionally, it was found that the matching the first-order moment is better than the cumulative probability, and that the data pairs comprising 30 to 100% are better for the PMM application. Finally, above findings were applied to a real rainfall event observed by the Bislsan radar and optimal parameters were estimated. The radar rain rate derived by applying these parameters was found to be well matched to the rain gauge rain rate.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.52 no.4
• /
• pp.175-183
• /
• 2015

The proposed algorithm consists of two stages: the learning and synthesis stages. At the learning stage, we first collect various high-resolution (HR)-low-resolution (LR) text image pairs, and quantize the LR images, and extract HR-LR block pairs. Based on quantized LR blocks, the LR-HR block pairs are clustered into a pre-determined number of classes. For each class, an optimal 2D-FIR filter is computed, and it is stored into a dictionary with the corresponding LR block for indexing. At the synthesis stage, each quantized LR block in an input LR image is compared with every LR block in the dictionary, and the FIR filter of the best-matched LR block is selected. Finally, a HR block is synthesized with the chosen filter, and a final HR image is produced. Also, in order to cope with noisy environment, we generate multiple dictionaries according to noise level at the learning stage. So, the dictionary corresponding to the noise level of the input image is chosen, and a final HR image is produced using the selected dictionary. Experimental results show that the proposed algorithm outperforms the previous works for noisy images as well as noise-free images.

• Journal of KIISE:Software and Applications
• /
• v.30 no.7_8
• /
• pp.672-680
• /
• 2003

A shopbot is a software agent whose goal is to maximize buyer´s satisfaction through automatically gathering the price and quality information of goods as well as the services from on-line sellers. In the response to shopbots´ activities, sellers on the Internet need the agents called pricebots that can help them maximize their own profits. In this paper we adopts Q-learning, one of the model-free reinforcement learning methods as a price-setting algorithm of pricebots. A Q-learned agent increases profitability and eliminates the cyclic price wars when compared with the agents using the myoptimal (myopically optimal) pricing strategy Q-teaming needs to select a sequence of state-action fairs for the convergence of Q-teaming. When the uniform random method in selecting state-action pairs is used, the number of accesses to the Q-tables to obtain the optimal Q-values is quite large. Therefore, it is not appropriate for universal on-line learning in a real world environment. This phenomenon occurs because the uniform random selection reflects the uncertainty of exploitation for the optimal policy. In this paper, we propose a Mixed Nonstationary Policy (MNP), which consists of both the auxiliary Markov process and the original Markov process. MNP tries to keep balance of exploration and exploitation in reinforcement learning. Our experiment results show that the Q-learning agent using MNP converges to the optimal Q-values about 2.6 time faster than the uniform random selection on the average.

• Proceedings of the Korea Inteligent Information System Society Conference
• /
• 2001.01a
• /
• pp.59-64
• /
• 2001

The robot soccer simulation game is a dynamic multi-agent environment. In this paper we suggest a new reinforcement learning approach to each agent's dynamic positioning in such dynamic environment. Reinforcement learning is the machine learning in which an agent learns from indirect, delayed reward an optimal policy to chose sequences of actions that produce the greatest cumulative reward. Therefore the reinforcement learning is different from supervised learning in the sense that there is no presentation of input pairs as training examples. Furthermore, model-free reinforcement learning algorithms like Q-learning do not require defining or learning any models of the surrounding environment. Nevertheless it can learn the optimal policy if the agent can visit every state- action pair infinitely. However, the biggest problem of monolithic reinforcement learning is that its straightforward applications do not successfully scale up to more complex environments due to the intractable large space of states. In order to address this problem. we suggest Adaptive Mediation-based Modular Q-Learning (AMMQL)as an improvement of the existing Modular Q-Learning (MQL). While simple modular Q-learning combines the results from each learning module in a fixed way, AMMQL combines them in a more flexible way by assigning different weight to each module according to its contribution to rewards. Therefore in addition to resolving the problem of large state effectively, AMMQL can show higher adaptability to environmental changes than pure MQL. This paper introduces the concept of AMMQL and presents details of its application into dynamic positioning of robot soccer agents.

• Journal of Internet Computing and Services
• /
• v.5 no.4
• /
• pp.53-62
• /
• 2004

This paper presents a study on the optimization for brokering between cargos and ships for future e-logistics. The primary contribution of this research is that we establish an optimization model by formalizing the criteria for the brokering such as time constraints, weight constraints, and preference values between cargos and ships. Another important contribution is that we not only investigate the complexity and the tractability of the optimal brokering problem but present how to evaluate the performance of the optimization program through an experiment. We first derive the preference values between cargos and ships using the time and the weight constraints. These preference values between each pair of cargos and ships are assigned to corresponding binary decision variables as coefficients in the objective function. The optimization model selects pairs of cargos and ships in a way that the sum of the preference values is maximized while satisfying given constraints. Experiment shows that the Davis-Putnam based optimization program finds optimal solutions in reasonable time for the problems with less than 90 decision variables.

• Proceedings of the Korea Society for Simulation Conference
• /
• 2001.10a
• /
• pp.321-324
• /
• 2001

The robot soccer simulation game is a dynamic multi-agent environment. In this paper we suggest a new reinforcement learning approach to each agent's dynamic positioning in such dynamic environment. Reinforcement Beaming is the machine learning in which an agent learns from indirect, delayed reward an optimal policy to choose sequences of actions that produce the greatest cumulative reward. Therefore the reinforcement loaming is different from supervised teaming in the sense that there is no presentation of input-output pairs as training examples. Furthermore, model-free reinforcement loaming algorithms like Q-learning do not require defining or loaming any models of the surrounding environment. Nevertheless it can learn the optimal policy if the agent can visit every state-action pair infinitely. However, the biggest problem of monolithic reinforcement learning is that its straightforward applications do not successfully scale up to more complex environments due to the intractable large space of states. In order to address this problem, we suggest Adaptive Mediation-based Modular Q-Learning(AMMQL) as an improvement of the existing Modular Q-Learning(MQL). While simple modular Q-learning combines the results from each learning module in a fixed way, AMMQL combines them in a more flexible way by assigning different weight to each module according to its contribution to rewards. Therefore in addition to resolving the problem of large state space effectively, AMMQL can show higher adaptability to environmental changes than pure MQL. This paper introduces the concept of AMMQL and presents details of its application into dynamic positioning of robot soccer agents.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.15 no.1
• /
• pp.1-5
• /
• 2005

The purpose of reinforcement learning is to maximize rewards from environment, and reinforcement learning agents learn by interacting with external environment through trial and error. Q-Learning, a representative reinforcement learning algorithm, is a type of TD-learning that exploits difference in suitability according to the change of time in learning. The method obtains the optimal policy through repeated experience of evaluation of all state-action pairs in the state space. This study chose n-Queen problem as an example, to which we apply reinforcement learning, and used Q-Learning as a problem solving algorithm. This study compared the proposed method using reinforcement learning with existing methods for solving n-Queen problem and found that the proposed method improves the convergence rate to the optimal solution by reducing the number of state transitions to reach the goal.