• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.12 no.1
• /
• pp.36-41
• /
• 2012

We tested a single pendulum simulation and observed the influence of several situation space segmentation types in reinforcement learning processes in order to propose a new adaptive automation for situation space segmentation. Its segmentation is performed by the Contraction Algorithm and the Cell Division Approach. Also, its automation is performed by "entropy," which is defined on action values’ distributions. Simulation results were shown to demonstrate the influence and adaptability of the proposed method.

• Journal of Service Research and Studies
• /
• v.7 no.3
• /
• pp.37-55
• /
• 2017

Recently, many business organizations have been increasingly expanding the utilization of the intelligent chatbot services for effective customer relation management. The purpose of this study is to empirically investigate the factors which affect the use of the intelligent chatbot services. Above all, the research model and the hypotheses were derived through reviewing the major relevant theories such as UX(user experience) theory(Honeycomb model), TRA(theory of reasoned action), TAM(technology acceptance model) and ETAM(extended TAM). And then, structural equation analyses using SmartPLS 3.0 for 233 valid questionnaires replies collected through the field survey was performed to test the hypotheses. Theoretically, this study can contribute to providing the conceptual framework with regard to enhancing the use of intelligent chatbot services. And practically, this study sheds new light on suggesting the guidelines to designing the UX(user experience) of the intelligent chatbot services.

• 제어로봇시스템학회:학술대회논문집
• /
• 2003.10a
• /
• pp.2027-2032
• /
• 2003

For action selection as well as learning, simple associations between stimulus and response have been employed in most of literatures. But, for a successful task accomplishment, it is required that an animat can learn and express behavioral sequences. In this paper, we propose a novel action-selection-mechanism to deal with sequential behaviors. For this, we define behavioral motivation as a primitive node for action selection, and then hierarchically construct a network with behavioral motivations. The vertical path of the network represents behavioral sequences. Here, such a tree for our proposed ASM can be newly generated and/or updated, whenever a new sequential behaviors is learned. To show the validity of our proposed ASM, three 2-D grid world simulations will be illustrated.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.5
• /
• pp.621-626
• /
• 2003

In this paper, we modeled a virtual life (VL) that reacts to the user s action according to its own behavioral characteristics and grows itself. We established some conditions with which such a VL is designed. Genetic Algorithm is used for the growth process that changes the VL s properties. In this process, the parameter values of the VL s properties are encoded as one chromosome, and the GA operations change this chromosome. The VL s reaction to the user s action is determined by these properties as well as the general expectation of each reaction. These properties are evaluated through 5 fitness measures so as to deal with multi-objective criteria. Here, we present the simulation of the growth Process, and show some experimental results.

• Journal of Korea Multimedia Society
• /
• v.25 no.6
• /
• pp.807-818
• /
• 2022

In order to further improve the accuracy and time efficiency of behavior recognition in intelligent monitoring scenarios, a human behavior recognition algorithm based on YOLO combined with LSTM and CNN is proposed. Using the real-time nature of YOLO target detection, firstly, the specific behavior in the surveillance video is detected in real time, and the depth feature extraction is performed after obtaining the target size, location and other information; Then, remove noise data from irrelevant areas in the image; Finally, combined with LSTM modeling and processing time series, the final behavior discrimination is made for the behavior action sequence in the surveillance video. Experiments in the MSR and KTH datasets show that the average recognition rate of each behavior reaches 98.42% and 96.6%, and the average recognition speed reaches 210ms and 220ms. The method in this paper has a good effect on the intelligence behavior recognition.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09a
• /
• pp.690-693
• /
• 2003

The purpose of two player game is that a player beats an enemy. In order to win to various enemies, a learning of various strategies is indispensable. However, the optimal action to overcome the enemies will change when the game done over and again because the enemy's actions also change dynamically. Sol it is din-cult that the player aquires the optimal action and that the specific player keeps winning to various enemies. Species who have a competition relation and affect other's existence is called a coevolution. Coevolution has recently attracred considerable interest in the community of Artificial Life and Evolutionary Computation(1). In this paper, we apply Classifier System for agent team to two player game. A reward and a penalty are given to the used rules when the agent achieve specific action in the game and each team's rulebase are evaluated based on the ranking in the league. We show that all teams can acquire the optimal actions by coevolution.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1997.11a
• /
• pp.98-103
• /
• 1997

Defuzzification plays a great role in fuzzy control system. Defuzzification is a process which maps from a space defined over an output universe of discourse into a space of nonfuzzy(crisp) number. But, it's impossible to convert a fuzzy set into a numeric value without losing some information during defuzzification. Also it's very hard to find a number that best represents a fuzzy set. Many methods have been used for defuzzification but most of then were problem dependent. There has been no rule which guides how to select a method that is suitable to solve given problem. Here, we have investigated most widely used methods and we have analyzed their characteristics and evaluated them. D. Driankov and Mizumoto have suggested 5 criteria which the‘ideal’defuzzification method should satisfy. But, they didn't considered about control action. Output fuzzy set if not only a fuzzy set but also a sequence of control action. We suggested 4 new criteria which describe sequence of cont ol action from some experiments. In addition, we have compared each method in simple adaptive fuzzy control. COG(Center of Gravity), or COS(Center of Sums) methods were successful in fuzzy control. However, at transition region, MOM(Mean of Maxima) was best among others in adaptive fuzzy control.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2000.05a
• /
• pp.245-248
• /
• 2000

In this paper we improve the performance of autonomous mobile robot by induction of reinforcement learning concept. Generally, the system used in this paper is divided into two part. Namely, one is neural-fuzzy and the other is dynamic recurrent neural networks. Neural-fuzzy determines the next action of robot. Also, the neural-fuzzy is determined to optimal action internal reinforcement from dynamic recurrent neural network. Dynamic recurrent neural network evaluated to determine action of neural-fuzzy by external reinforcement signal from environment, Besides, dynamic recurrent neural network weight determined to internal reinforcement signal value is evolved by genetic algorithms. The architecture of propose system is applied to the computer simulations on controlling autonomous mobile robot.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.05a
• /
• pp.325-329
• /
• 2003

This paper discuss the method of the system's efficient control using a Intelligent hybrid algorithm in nonlinear dynamics systems. Existing neural network and genetic algorithm for the control of non-linear systems work well in static states. but it be not particularly good in changeable states and must re-learn for the control of the system in the changed state. This time spend a lot of time. For the solution of this problem we suggest the intelligent hybrid self-tuning controller. it includes neural network, genetic algorithm and immune system. it is based on neural network, and immune system and genetic algorithm are added against a changed factor. We will call a change factor an antigen. When an antigen broke out, immune system come into action and genetic algorithm search an antibody. So the system is controled more stably and rapidly. Moreover, The Genetic algorithm use the memory address of the immune bank as a genetic factor. So it brings an advantage which the realization of a hardware easy.

• Journal of Korea Game Society
• /
• v.9 no.4
• /
• pp.81-88
• /
• 2009

In this paper, I suggested a way for searching a path of the intelligent character in an action game by using a genetic algorithm. This realized the algorithm which enables not only to chose the nearest path but also to search the optimum path by using genetic algorithm. In this case, if the codes of chromosomes are applied as they are, a lot of lethal genes could occur. In order to solve such a problem, I used a splicing method, one of the DNA's behavior characteristics. The intelligent character searched out a optimum pass as well as a shortcut path with one treatment by using the characteristic of a genetic algorithm which generates multiple candidate solutions in the search process.