We propose a method of association rule mining using genetic network programming (GNP) with time series processing mechanism and attribute accumulation mechanism in order to find time related sequence rules efficiently in association rule extraction systems. We suppose that, the database consists of a large number of attributes based on time series. In order to deal with databases which have a large number of attributes, GNP individual accumulates better attributes in it gradually round by round, and the rules of each round are stored in the Small Rule Pool using hash method, and the new rules will be finally stored in the Big Rule Pool. The aim of this paper is to better handle association rule extraction of the database in many time-related applications especially in the traffic prediction problem. In this paper, the algorithm capable of finding the important time related association rules is described and experimental results considering a traffic prediction problem are presented.
The auction mechanism is widely used in Web-based sites and originally designed for human beings, but it might not be the most efficient one in the future, while, there is a demand of evolutionary computation auction agents adaptable to the dynamic auction environments. In this paper, we have applied genetic network programming (GNP) to auction agents to determine a bid at each time step and developed multiple round English auction mechanisms based on multi-agent systems. In the simulations, we provide comparisons of the proposed method with existing ones. As a result, it has been found that the proposed method could help agents to evolve their strategies generation by generation to get more goods with less money. Also, GNP has a good performance of helping the agent to find out the suitable strategy under the current situation.
As an extension of Genetic Algorithm (GA) and Genetic Programming (GP), a new approach named Genetic Network Programming (GNP) has been proposed in the evolutionary computation field. GNP uses multiple reusable nodes to construct directed-graph structures to represent its solutions. Recently, many research has clarified that GNP can work well in data mining area. In this paper, a novel evolutionary paradigm named GNP with Estimation of Distribution Algorithms (GNP-EDAs) is proposed and used to solve traffic prediction problems using class association rule mining. In GNP-EDAs, a probabilistic model is constructed by estimating the probability distribution from the selected elite individuals of the previous generation to replace the conventional genetic operators, such as crossover and mutation. The probabilistic model is capable of enhancing the evolution to achieve the ultimate objective. In this paper, two methods are proposed based on extracting the probabilistic information on the node connections and node transitions of GNP-EDAs to construct the probabilistic model. A comparative study of the proposed paradigm and the conventional GNP is made to solve the traffic prediction problems using class association rule mining. The simulation results showed that GNP-EDAs can extract the class association rules more effectively, when the number of the candidate class association rules increases. And the classification accuracy of the proposed method shows good results in traffic prediction systems.
Recently, Artificial Intelligence (AI) technology has been applied to many applications. As an extension of Genetic Algorithm (GA) and Genetic Programming (GP), Genetic Network Programming (GNP) has been proposed, whose gene is constructed by directed graphs. GNP can perform a global searching, but its evolving speed is not so high and its optimal solution is hard to obtain in some cases because of the lack of the exploitation ability of it. To alleviate this difficulty, we developed a hybrid algorithm that combines Genetic Network Programming (GNP) with Ant Colony Optimization (ACO). Our goal is to introduce more exploitation mechanism into GNP. In this paper, we applied the proposed hybrid algorithm to a complicated real world problem, that is, Elevator Group Supervisory Control System (EGSCS). The simulation results showed the effectiveness of the proposed algorithm.
In this paper, we propose a heuristic method trying to find a good approximation to the global optimum route for origin-destination pairs through iterations until the total traveling time converges in static traffic systems. The overall idea of our method is to iteratively update the traveling time of each route section according to its corresponding traffic volume, and continuously generate a new global route by Q value-based dynamic programming combined with Boltzmann distribution. Finally, we can get the global optimum route considering the traffic volumes of the road sections. The new proposed method is compared with the conventional shortest-path method and the result demonstrates that the proposed method performs better than the conventional method in global perspective.
We present a new method to accelerate the HITS algorithm by exploiting hyperlink structure of the web graph. The proposed algorithm extends the idea of authority and hub scores from HITS by introducing two diagonal matrices which contain constants that act as weights to make authority pages more authoritative and hub pages more hubby. This method works because in the web graph good authorities are pointed to by good hubs and good hubs point to good authorities. Consequently, these pages will collect their scores faster under the proposed algorithm than under the standard HITS. We show that the authority and hub vectors of the proposed algorithm exist but are not necessarily be unique, and then give a treatment to ensure the uniqueness property of the vectors. The experimental results show that the proposed algorithm can improve HITS computations, especially for back button datasets. Comment: 10 pages, 3 figures, to be appear in Journal of Advanced Computational Intelligence and Intelligent Informatics, Vol. 14 No. 1, 2010
A mechanism called Eligibility Propagation is proposed to speed up the Time Hopping technique used for faster Reinforcement Learning in simulations. Eligibility Propagation provides for Time Hopping similar abilities to what eligibility traces provide for conventional Reinforcement Learning. It propagates values from one state to all of its temporal predecessors using a state transitions graph. Experiments on a simulated biped crawling robot confirm that Eligibility Propagation accelerates the learning process more than 3 times.
The access control and scalable encryption scheme we propose for JPEG 2000 encoded images encrypts JEPG 2000 codestreams using the SNOW 2 progressive encryption algorithm to encrypt resolutions, quality layers, or packets independently to provide resolution, quality or fine-grain scalability. Access is controlled to different image resolutions or quality levels granted to different users receiving the same encrypted JPEG 2000 codestream but having different decryption keys. Keys used with successive resolutions or quality layers are mutually dependent based on the SHA-256 one-way hashing function. Encrypted JPEG 2000 codestreams are transcoded by an intermediate untrusted network transcoder, without decryption and without access to decryption keys. Our encryption scheme preserves most of the inherent flexibility of JPEG 2000 encoded images and is carefully designed to produce encrypted codestreams backward compatible with JPEG 2000 compliant decoders.
We discuss intelligent information provision involving different modal information collaboratively presented, with an example of news articles about stock prices summarized based on 2D chart representation on stock prices. We use the MuST corpus, an annotated corpus for easily extracting trends in information, e.g., statistical values, etc., as the news article corpus to be summarized. We associate the MuST corpus with numerical data on the stock prices, and propose a way to provide people with a summarized text about news articles on prices corresponding to 2D chart representation.
A 3D (three-dimensional) pseudo-reconstruction method from a single image is presented as a novel approach reconstructing a 3D model with no prior internal knowledge of outdoors image. In the proposed method, an image is represented as a collection of sky layer, ground layer, and object layer. A visual radical coordinate system with vanishing point is established to accommodate the extracted 3D data from images. Learning method is done via the layers database. The experiment results show that the visually acceptable 3D model can be extracted less than one minute. That means a higher resolution in much shorter time, compared to conventional methods. This method can be applied to computer games, industrial measurement, archeology, architecture and visual realities.
A local character tensor is proposed for the automatic three-dimensional (3D) pair-wise registration based on free-view 3D datasets. In the proposed method, there are two characters, i.e., the optimal segmentation to realize the automatic processing and local character tensor to improve the matching probability. It is applied for solving the mismatching problem and large-scale 3D datasets, using non-structured datasets are tested in a PC with Intel Pentium M 1.50 GHz and 1.0 GB memory. Pair-wised experimental results show the proposed method increases average 12.6% matching probability and decreases average 18.9 seconds computational time compared to the conventional local character based registration method. This registration method can be further applied to 3D reconstruction from navigation, model based object recognition to accurate 3D geometric object model application.
Knowledge based systems need to deal with aggrega- tion and fusion of data with uncertainty. To use many sources of information in numerical forms for the pur- pose of decision or conclusion, systems suppose to have tools able to represent the knowledge in a mathemat- ical form. One of the solutions is to use fuzzy logic operators. We present in this article an improvement of the triple Π operator introduced by Yager and Ry- balov, which is called mean 3Π. Whereas triple Π is an operator completely reinforced, the presented opera- tor is a mean operator, which makes it more robust to noise. to simplify calculation. Consequently, the fit of the model to the expert opinion is debatable. An expert prefers to provide intervals rather than isolated values because his knowledge is of reliability limited and spoilt by inaccu- racy. Belief theory handles inaccurate and uncertain in- formation. The possibility theory and fuzzy set are based on fuzzy logics. Fuzzy logics are characterized as "logic based on the real number". In these types of logic con- sider that the truth degree are taken from the real line R. Fuzzy modeling and control are typical examples of tech- niques that make use of human knowledge and deductive process basically using inference mechanisms. The ad- vantage of fuzzy modeling is that the information can be either of numerical or of symbolic nature. Its representa- tion as numerical degrees leads to a quantification of its characteristics (uncertain, imprecise, incomplete) which have to be taken into account in a fusion process. There- fore, the kernel of these mechanisms is the fusion opera- tors defined as:
A classification algorithm for abdominal organs in ultrasonic test images based on the operator's knowledge is proposed. This is in order to use the medical images included in medical charts for secondary uses, e.g., medical data analysis. It makes a correlation between target organs in test images and search unit information on the body mark region. In the central region of abdominal images, target organs are uniquely determined through recognition of the liver region and in consideration of the location of the diaphragm. A classification experiment, done using 600,000 real test images taken at the Kochi Medical School Hospital from 2004 to 2008, was carried out to evaluate the performance of the proposed system in terms of accuracy rate of detection of the body mark region and diaphragm region. The proposed algorithm constitutes an essential classification system for the secondary use of a large database of ultrasound images taken in the course of medical practice.
Recent results in retinal research have shown that ganglion cell receptive fields cover the mammalian retina in a mosaic arrangement, with insignificant amounts of overlap in the central fovea. This finding implies that from the informatics point of view there is a major conceptual gap between traditional and widely accepted, convolution based image filtering algorithms, and the way visual information is processed by the retina in the eye. The use of traditional filters with non-overlapping operator architectures leads to considerable information losses between centers of filter kernels. This paper introduces a novel model of the eye-retina system that fills the conceptual gap of information processing between the retina and the overlapping (convolution based) architectures used by today's widely adapted algorithms. The proposed computational model takes into consideration data convergence, as well as the dynamic and optical properties of the eye lens. Based on the evaluation of the model, three hypotheses are formulated on the role of the optical precision of the eye-lens and involuntary eye accommodation dynamics.
The method for generating abstract images from a set of images is proposed. The method selects a representative image from a given set of images, in which the common features in terms of the composition are highlighted with image processing techniques. Common features are extracted based on Local Similarity Pattern (LSP), which has been originally proposed for image retrieval. The selection of representative images is performed based on the difference between the color histogram calculated from a set of regions, of which color features are common, and that calculated from the remaining regions. The experimental results show the performance of the proposed method, in terms of its effectiveness for image classification, as well as the accuracy of selecting representative images. The concept of abstract images is expected to be useful for developing a directory service for searching images on the Web.
This paper investigates the characteristics of robots for non-industrial use such as home robots, when those are used as an interface for accessing information. Although information support is one of important capabilities home robots should have, the merits of accessing information via a robot compared with the access via PC or a mobile phone have yet to be fully explored. This paper focuses on the physical presence of robots, which is supposed to be important for robots to provide users with information. In order to investigate the merits, two experiments with participants are performed in this paper. The main contributions of the paper are the following points. First, it is shown that a robot can effectively attract the participants through movements, even though they pay less attention to it. Second, the possibility of using robot actions for providing additional information about information to be accessed by the participant is also investigated. Finally, the effect of a robot User Interface (UI) prototype on communication among users when providing information to them is also investigated. The obtained results support the significance of information support by home robots, which will be used for designing home robots with information support facility.
Although wireless LAN is useful in its small size and mobility, the connection region of transmitted radio wave is strongly affected by other electric devices, consumer products, and differences in size and type of the room. Besides, wireless LAN points (APs) must exclude a personal computers without permitting to connect to the Internet. Therefore, how optimally APs are located is important. In this paper, we propose the APs' optimal location method. The proposed algorithm integrates fuzzy rules acquired by fuzzy ID3 with knowledge of security experts, and estimates the connection region for AP. We discuss how to formulate the method for setting AP optimal location and show the effectiveness of this method by illustrating the examples of AP optimal locations.
Wireless access is needed to serve rural and urban sectors in the developing nations, while simultaneously encompassing both data and voice traffic in the same core network. We propose an architectural design of Multiservice Provisioning Platform (MSPP) for WiMAX over SONET/SDH, which involves interworking conventional technologies to deliver a solution that caters to unique situations that favor wireless access network over fully cabled systems. It consists of Wireless Access Network design, 2-way interlinking section, and the MSPP – optical core Network. It enhances the versatility of the MSPP and uses the SONET/SDH core network efficiently.
This paper deals with the problem of chaotic disturbances accommodation when these are generated by known non linear dynamics. In order to accomplish this goal, Takagi-Sugeno fuzzy models are called for as they offer the advantage of having virtually a linear rule consequent to approximate non linear systems. A control law inspired from the known disturbance accommodation control theory (DAC theory) is used to make the effects of disturbances vanish or attenuated while the considered linear plant is stabilized at the same time. An illustrative example is provided.
Personal authentication is becoming an increasingly important problem. Online signature verification is one promising form of biometric authentication. However, the verification accuracy of online signature verification is not high enough and still needs to be improved. To do so, we previously proposed a usergeneric fusion model. Although the verification accuracy was reasonably good, the proposed method cannot adequately take into account users' individuality. In this paper, to further improve the verification accuracy, we propose a method that can take such individuality into account. First, in a training phase, we divide a training dataset into several groups. Then, several fusion models are generated using a parameterized family of nonlinear functions by a Markov chain Monte Carlo method. In an enrollment phase, in order to take into account users' individuality, we introduce model reliability. The model reliability for each user and each model is different, enabling us to take users' individuality into account. In a verification phase, a marginal likelihood is calculated for each group. Then, a verification score is calculated by a weighted sum of the marginal likelihoods from the groups, using the model reliability. To evaluate the performance of the proposed algorithm, we conducted experiments using the SVC2004 database. The verification accuracy was improved over the previous algorithm.
Genetic Network Programming (GNP) is an evolutionary algorithm derived from GA and GP. Directed graph structure, reusability of nodes, and implicit memory function enable GNP to deal with complex problems in dynamic environments efficiently and effectively, as many paper demonstrated. This paper proposed a new method to optimize GNP by extracting and using rules. The basic idea of GNP with Rule Accumulation (GNP with RA) is to extract rules with higher fitness values from the elite individuals and store them in the pool every generation. A rule is defined as a sequence of successive judgment results and a processing node, which represent the good experiences of the past behavior. As a result, the rule pool serves as an experience set of GNP obtained in the evolutionary process. By extracting the rules during the evolutionary period and then matching them with the situations of the environment, we could, for example, guide agents' behavior properly and get better performance of the agents. In this paper, we apply GNP with RA to the problem of determining agents' behaviors in the Tile-world environment in order to evaluate its effectiveness. The simulation results demonstrate that GNP with RA could have better performances than the conventional GNP both in the average fitness value and its stability.
This paper aims at investigating acoustic features, which can objectively explain breathiness and roughness of elderly speech, respectively. In this paper, acoustic analysis was carried out using word sequences, which were uttered by 153 male speakers in the age range of between 20 and 89 years old. Concerning the breathiness, we confirmed that elderly breathy voices caused energy lift in higher frequency region over 4 kHz in average power spectra during the stationary parts in the uttered vowels. Concerning roughness, we observed the slight fluctuations, which synchronized with vocal cord vibration, in amplitude spectra during stationary parts of vowels. Based on acoustic analysis results, we propose physical parameters for measuring breathiness and roughness, respectively. In this paper, listening tests were carried out to quantitatively give the subject degrees of breathiness and roughness, respectively. It was confirmed that the proposed physical parameters had correlation with each of subjective degrees. Relationships between age and acoustic characteristics of breathiness and roughness were investigated using the proposed parameters. It is confirmed that the degree of breathiness and roughness increased in proportion to age, especially in age ranges over 60 years old.
This paper proposes using reinforcement learning to acquire a government bond trading strategy. We applied this method to the 10-year Japanese government bond (JGB) market and confirmed that it acquires profitable trading even in extrapolation.
A method of association rule mining with chi-squared test using Alternate Genetic Network Programming (aGNP) is proposed. GNP is one of the evolutionary optimization techniques, which uses directed graph structures as genes. aGNP is an extended GNP in terms of including two kinds of node function sets. The proposed system can extract important association rules whose antecedent and consequent are composed of the attributes of each family defined by users. Rule extraction is done without identifying frequent itemsets used in Apriori-like methods. The method can be applied to rule extraction from dense database, and can extract dependent pairs of the sets of attributes in the database. Extracted rules are stored in a pool all together through generations and reflected in genetic operators as acquired information. In this paper, we describe the algorithm capable of finding the important association rules and present some experimental results.
Knowledge- and sample-based learning approaches play a pivotal role in image processing. However, the acquisition and integration of expert knowledge (for the former) and providing a sufficiently large number of training samples (for the latter) are generally hard to perform and time-consuming tasks. Hence, learning image processing tasks from a few gold/ground-truth samples, prepared by the user, is highly desirable. This paper demonstrates how the combination of an optimizer (e.g., genetic algorithm) and image processing tools (e.g., parameterized morphology operations) can be used to generate image processing procedures for image filtering and object extraction. For this purpose, the approach receives the original and the user-prepared image (filtered image or image with extracted target object) as a gold sample which reflects the user's expectations. After carrying out the training or optimization phase, the optimal procedure is generated and ready to be applied to new images. The feasibility of our approach is investigated for two individual image processing categories, namely filtering and object extraction, by well-prepared synthetic images. The proposed architecture and the employed methodologies are explained in detail. Experimental results are provided as well.
The subject matter in this work is covered by a US provisional patent application.
We focus on robotic learning under multiple instructors. Even when their goal is the same, different instructors inevitably was different approaches. We propose incorporating DP matching and clustering, classifying the teaching demonstrations of instructors into groups of similar ones. Experiments in which an AIBO robot was taught to walk forward demonstrated that our proposal acquired appropriate teaching approaches based on AIBO’s different embodiments and maximizing task accomplishment.
In multi-robot system, cooperation is needed to execute tasks efficiently. The purpose of this study is to realize cooperation among multiple robots using in- teractive communication. An important role of communication,in multi-robot sys- tem is to make,it possible to control other robots by intention transmission. We consider that multi-robot system can be more and more adaptive by treating com- munication as action. In this report, we adopt action adjustment function to achieve cooperation between robots. We also run some computer,simulations of collision avoidance as an example of cooperative task, and discuss the results. Keywords.Q-learning, Multi-Robot System, Communication, Cooperation, Mobile Robot
Enabling a digital actor to move autonomously in a virtual environment is a challenging problem that has attracted much attention in recent years. The systems proposed in several researches have been able to plan the walking motions of a humanoid on an uneven ter- rain. In this paper, we aim to design a planning sys- tem that can generate various types of motions for a humanoid with a unified planning approach. Based on our previous work, we add two additional motion abil- ities: leaping and moving obstacles into the system. In previous work, the order of moving obstacles is deter- mined first, and then the corresponding paths for the pushing/pulling motions are generated. In this work, we take a unified approach that accounts for all types of motions at the same time. We have implemented a planning system with this unified approach for a hu- manoid moving in a layered virtual environment. Sev- eral simulation examples are demonstrated in this pa- per to illustrate the effectiveness of the system.
We discuss the realization of rapid movement for legged entertainment robots using two new actuators, the Inertia Actuator and the Cam Charger. As an internal torque generator, the Inertia Actuator generates small internal torque by changing the rotor speed and large internal torque quickly by using a brake to stop the rotor at high speed. To realize jumping, we introduce the Cam Charger to fit to the robot foot. The key is to charge a series of strong torsion springs using a specially shaped cam. After detailing of the Cam Charger and the Inertia Actuator principles, we evaluate the feasibility of our approach through simulation. We show experimentally that our artistic “Jumping Joe” robot prototype including these two actuators demonstrates rapid movements such as fast wakeup, jumping, and somersaults.
This paper proposes fuzzy compensation for actuators\' motion forces, (dynamics, gravity and friction) in a force/motion control algorithm for the assembly of segments of a shield tunnel excavation applying a 6 DOF hydraulic parallel link manipulator. First, we introduce the feedback force/motion control algorithm with fuzzy forces compensation. Then, we introduce a rule-base fuzzy compensating model and its real-time implementation for every hydraulic actuator of a Stewart Platform so that we can reduce the effect of friction forces and hence improve the quality of force control and assembly. The experimental results of the control system with and without fuzzy compensation are presented, which show a good achievement in contact force estimation and manipulator motion utilizing the proposed fuzzy compensation
Adapting real mobile robots to complex or dynamic environments is just one of the many challenges robotics researchers face. The difficulty in such environments is in developing a simple, quick adaptive controller that adapts robots to patterns in these environments, especially when individual patterns require unique behavior from the robot. Although most standard evolutionary algorithms attempt to obtain optimal networks for such environments, this is difficult to attain due to network confusion in adapting and readapting patterns. We propose a simple adaptive controller able to learn and remember. It simplifies environments into simple groups of patterns, each of which the robot can independently learn and memorize. The memory introduced in the controller enhances the robot's ability to track its own experience and to cope with upcoming events. Experimental results show that the controller handles general complexity and gives the robot more adaptability, stability, and autonomy.
The never unseen information explosion in data transmission and communication called for new methods in signal coding and reconstruction. To minimize the channel capacity needed for the transmission urged researchers to find techniques which are flexible and can adapt to the available space and time. Anytime techniques are good candidates for such purposes. If the signal/data to be transmitted can be characterized as sequence of stationary intervals overcomplete signal representations can be applied. These techniques can be operated in an anytime manner as well, i.e., are excellent tools for handling the capacity problems. This paper introduces the concept of anytime recursive overcomplete signal representations using different recursive signal processing algorithms. The novelty of the approach is that an on-going set of signal transformations together with appropriate (e.g., L 1 norm) minimization procedures can provide optimal and flexible anytime on-going representations, ongoing signal segmentations into stationary intervals, and on-going feature extractions for immediate utilization in data transmission, communication, diagnostics, or other applications. The proposed technique may be advantageous if the transmission channel is overloaded and in case of processing non-stationary signals when complete signal representations can be used only with serious limitations because of their relative weakness in adaptive matching of signal structures.
A novel optimization method named RasID-GA (an abbreviation of Adaptive Random Search with Intensification and Diversification combined with Genetic Algorithm) is proposed in order to enhance the searching ability of conventional RasID, which is a kind of Random Search with Intensification and Diversification. In this paper, the timing of switching from RasID to GA, or from GA to RasID is also studied. RasID-GA is compared with parallel RasIDs and GA using 23 different objective functions, and it turns out that RasID-GA performs well compared with other methods.
This paper introduces a componentwise joint receipt operation ⊕ on an n -component product set Π n i=1 g i , and develops an axiom system to justify an additive representation for a binary relation ≿ on Π n i=1 g i . Basically, our axiom system is similar to the n -component ( n ≥ 3), additive conjoint structure. However, the introduction of the operation ⊕ yields two new axioms – additive solvability and invariance under multiplication – and hence we can weaken the independence axiom of the conjoint structure. The weakened independence axiom requires the independence of the order for each single factor from fixed levels of the other factors, while the conjoint structure involves the independence of the order for two or more factors. Finally, it is shown by a brief experimental test that the weakened independence axiom is sustained.
To check whether a new algorithm is better, re- searchers use traditional statistical techniques for hypotheses testing. In particular, when the results are inconclusive, they run more and more simulations (n2 > n1, n3 > n2, . . . , nm > nm¡1) until the results become conclusive. In this paper, we point out that these results may be misleading: in the traditional approach, we select a statistic and then choose a threshold for which the probability of this statistic "accidentally" exceeding this threshold is smaller than, say, 1%. It is very easy to run additional simulations with ever-larger n. The probability of error is still 1% for each ni, but the probability that we reach an erroneous conclusion for at least one of the values ni increases as k increases. In this paper, we design new statistical techniques oriented towards experiments on simulated data, techniques that would guarantee that the error stays under, say, 1% no matter how many experiments we run.
In this research we design a network administrator assistance system based on traffic measurement and fuzzy c-means (FCM) clustering analysis method. Network traffic measurement is an essential tool for monitoring and controlling communication network. It can provide valuable information about network traffic-load patterns and performances. The proposed system utilizes the FCM method to analyze users' network behaviors and traffic-load patterns based on traffic measurement data of IP network. Analysis results can be used as assistance for administrator to determine efficient controlling and configuring parameters of network management systems. The system is applied in Dali University campus network, and it is effective in practice.
Potential applicants to graduate school find it dicult to predict, even approximately, which schools will accept them. We have created a predictive model of admissions decision- making, packaged in the form of a web page that allows students to enter their information and see a list of schools where they are likely to be accepted. This paper explains the rationale for the model's design and parameter values. Interesting issues include the way that evidence is combined, the estimation of parameters, and the modeling of uncertainty. Index Terms student assessment, acceptance criteria, application evaluation, decision-making, combination of evi- dence, ordered weighted average, uncertainty, GRE scores, GPA, letters of recommendation
In the real-world multiagent/multirobot problems, a position of each agent is an important factor to affect agents’ performance. In the real-world problem such as soccer, the agent(player)’s position should be changed based on the current environmental state. Because the real-world problem is generally dynamic and continuous, assigning the most desirable position for any state is not possible. We formalize this issue as a map from a focal point like a ball position in a soccer field to a desirable position of each agent. We conducted experiments showing that agent positioning are acquired efficiently through intuitive human operation and function approximation models using supervised learning, and conducted performance evaluation experiments to determine the most suitable model. In performance evaluation experiments, we evaluated the generalization capability for each model with datasets which several data are randomly removed from original dataset and dataset which several specific data are intentionally removed from original dataset. Experiment results showed that our proposed function approximation model combining Delaunay triangulation and linear interpolation produced the highest performance.
This paper focuses on developing human skills through interaction between a human player and a computer agent, and explores its strategic method through experiments on the bargaining games where human players negotiate with computer agents. Specifically, human players negotiate with three types of agents: (a) strong/weak attitude agents making aggressive/defensive proposals in advantageous/disadvantageous situations; (b) fair agents making fair proposals; and (c) the "human-like" agents making mutually agreeable proposals as the number of games increases. Analysis of the human subject experiments has revealed the three major implications: (1) human players negotiating with the strong/weak attitude agents obtain the largest profit overall; (2) human players negotiating with "human-like" agents win many games; and (3) no relationship exists between profit maximization and a win of the games.
In this study, the Coordinated Hybrid Agent (CHA) framework for the control of Multi-Agent Systems (MASs) is applied for a heterogeneous multi-agent system. The system consists of a mobile robot, an overhead crane, and a robot manipulator. The final goal for this project is to implement coordination tasks for the system. In this framework, the control of the MAS is regarded as a decentralized control and coordination of agents. A coordination rule base is developed for the intelligent coordination control layer. Experiments show that the framework is able to model the heterogeneous multi-agent systems.
This study focuses on gestures negotiation dialogs. Analyzing the situation/gesture relationship, we suggest how to enable agents to conduct adequate human-like gestures and evaluated whether an agent's gestures could give an impression similar to those by a human being. We collected negotiation dialogs to study common human gestures. We studied gesture frequency in different situations and extracted gestures with high frequency, making an agent gesture module based on the number of characteristics. Using a questionnaire, we evaluated the impressions of gestures by human users and agents, confirming that the agent expresses the same state of mind as the human being by generating an adequately human-like gesture.
This study attempts to apply an agent-based approach to modeling a transportation system. Utilizing the advantage of agent-based model of being validated at an individual level, a social dilemma situation of travel mode choice is modeled and viewed as a complex system. Inductive-learning's capability of travelers is used and combined with an evolutionary approach in order to simulate travelers' learning process. A user-equilibrium point as predicted by a conventional equilibrium analysis can be reached and stabilized. The stable situation is produced by interaction process among agents and by behavioral change process of each agent, without a central or external rule that organizes objective function of the system. The study also reveals some conditions that may produce other stable situations in addition to the user equilibrium point. An emergent situation combined with travelers' sensitivity to payoff differences is observed to be influential.
This paper presents a novel framework for modeling embodied conversational agent for crisis communication focusing on the H5N1 pandemic crisis. Our system aims to cope with the most challenging issue on the maintenance of an engaging while convincing conversation. What primarily distinguishes our system from other conversational agent systems is that the human-computer conversation takes place within the context of H5N1 pandemic crisis. A Crisis Communication Network, called CCNet, is established based on a novel algorithm incorporating natural language query and embodied conversation agent simultaneously. Another significant contribution of our work is the development of a Automated Knowledge Extraction Agent (AKEA) to capitalize on the tremendous amount of data that is now available online to support our experiments. What makes our system differs from typical conversational agents is the attempt to move away from strictly task-oriented dialogue.
We introduce 2-copulas (copulas, shortly) and recent related research results. We present invariant copulas and their application in the theory of aggregation operators. Copulas are transformed by increasing bijections at the unit interval and discuss copula attractors. We also present results on the approximation of associative copulas by strict and nilpotent triangular norms.
Analytic Hierarchy Process (AHP) is one of the most popular tools for supporting human decision making, and several fuzzy extensions of AHP have been proposed. The present study investigated psychological effects of both fuzzy ratings in fuzzy AHP and crisp feedback of the results from fuzzy AHP. The results suggest that fuzzy ratings could incorporate the fuzziness of a person’s feelings in his/her decision making. The results also suggest that crisp feedback, which exaggerates the superiority of only one alternative or the differences among the alternatives, could help a person to make his/her decision, especially when being deeply puzzled about his/her choice.
In many application areas, there is a need for interdisciplinary collaboration and education. However, such collaboration and education are not easy. On the example of our participation in a cyberinfrastructure project, we show that many obstacles on the path to successful collaboration and education can be overcome if we take into account that each person's knowledge of a statement is often a matter of degree - and that we can therefore use appropriate degree-based ideas and techniques.
In this paper we present the fuzzy description logic ALC_FH introduced, where primitive concepts are modified by means of hedges taken from hedge al- gebras. ALC_FH is strictly more expressive than Fuzzy-ALC defined in (11). We show that given a linearly ordered set of hedges primitive concepts can be modified to any desired degree by prefixing them with appropriate chains of hedges. Furthermore, we define a decision procedure for the unsatisfiability problem in ALC_FH , and discuss knowledge base expansion when using terminologies, truth bounds, expressivity as well as complexity issues. We extend (8) by allowing modifiers on non-primitive con- cepts and extending the satisfiability procedure to handle concept definitions.