Neuron dynamics, shown by an excitable membrane, were replicated using a two-terminal electronic circuit (Murai et al. 1977; Hoshimiya et al. 1979), which was composed of four diffierent kinds of current: corresponding to capacitance, leakage, sodium and potassium currents. The main purpose of this paper is to clarify that the electronic circuit is equivalent to the Hodgkin-Huxley equations, from a topological point of view. To this end, we introduce continuous differential equations reproducing bifurcation structures shown quantitatively by the electronic circuit. In the electronic circuit, the parameter E_K, the potassium equilibrium potentail proposed by authors, is responsible for reproducing bifurcation structures of the Hodgkin-Huxley equation.

抄録全体を表示

This paper discusses the agent system for negotiation support based on the fuzzy decision making. The membership function is a very important factor to raise the accuracy of fuzzy control. It can be created and tuned up in a systematic and autonomous way by appllying evolution system of genetic programming (GP). And, the effectiveness was examinde by applying to a practical negotiation case which needs coooperative decision making. As a result, the following findings were obtained. (1) It was possible to obtain the appropriate membership function suitable for the control object from among a vast search area comparatively easily. (2) The system was improved in respect of obtaining the negotiation result which offers mutual satisfactory compared to the system with a simple triangular membership function. (3) The result was more excellent compared to random search.

抄録全体を表示

Vanadyl-phtalocyanine (VOPc) thin films were prepared on polyethylene-terephthalate (PET) film. The VOPc film prepared on PET substrate of 120° had Phase II. The third harmonic (TH) intensity of the VOPc film was measured by Maker fringe. The third optical susceptibility (X⁽³⁾) estimated from TH intensity is 3.5×10⁹ esu. We also showed that the transition from Phase I to Phase II of the VOPc film was enhanced by corona charging.

抄録全体を表示

We constructed the simulation code of two-dimensional Photon Scanning Tunneling Microscope (2D-PSTM) based on integral equations called Guided-Mode Extracted Integral Equations. The code can treat the global-model of 2D-PSTM with a dielectric-probe and with a metal-coated dielectric-probe. Examples of simulations are shown for the case where the object is dielectric and metal. The simulation results shows that the interaction between the object and the probe is very complicated and that it is not easy to understand the physical process of 2D-PSTM intuitively.

抄録全体を表示

Optical flow, a two-dimensional (2D) motion field on image plane, is essential for such tasks as the visual guidance of locomotion through the environment, the manipulation and recognition of objects. However. recovering three-dimensional (3D) motion information from optical flow, is a difficult problem because the relationship between the optical flow field and 3D motion parameters of the observer along with the depth of the environment, is nonlinear. In this paper, we propose a new method for estimating 3D motion information from optical flow. Considering an observer moving through a static environment, we intend to recover observers 3D motion parameters and environment's relative depth map. Based on motion perspective, the estimation is carried out in three steps using three sets of equations derived from the nonlinear equation of motion perspective. First, direction of the translation components is recovered by searching a candidate over a discrete sampled space to minimize a residual function. Once the translation has been recovered. the rotation components of observer's 3D motion can be resolved from the second set of equations by using least square optimization. Finally. the estimation of relative depth map of the environment is straightforward using the third set of equations, given the recovered 3D motion parameters.

抄録全体を表示

abstract: In this paper, application of an eigenspace method to a visual manipulation system is described. In contrast to traditional visual control approaches, visual information captured by a CCD camera is directly processed in the proposed scheme, using the so called eigenspace method rather than shape recognition. A set of images of the controlled object obtained by being coarsely sampled from the workspace is compressed to a low-dimension space called as an eigenspace. Thus every image of the controlled object can be mapped to a point in the eigenspace. So, in another meaning, every state of the object corresponds to a point in the eigenspace. Moreover, a new method using a Radial Basis Functions Network to project the image's eigenspace coordination to object's state is developed. Thereby using a suitable eigenspace calculated in advance, a new image captured by the camera can be mapped into the eigenspace, and the corresponding object's state can be estimated during operation. As examples, a real-time visual manipulation system to perform the so-called “ball and beam” control task is adopted. Finally, performance of the scheme is examined by experiments.

抄録全体を表示

This paper considers the position tracking problem of a voltage-controlled magnetic levitation system. It is well known that the control problem is quite complicated and challenging due to inherent nonlinearities associated with the electromechanical dynamics. And feedback linearization approach via coordinate transformation is considered to be a useful candidate to tackle the control problem. Usually however, feedback linearization control does not guarantee exact linearization and robustness in the presence of model uncertainties. In this paper. we propose a robust nonlinear controller in the presence of parametric uncertainties. The design procedure is carried out in a backstepping design manner, by employing nonlinear damping terms to suppress the effects of parametric uncertainties which may cause instability. Input-to-state stable property of the control system is analyzed, and experimental results are included to show the excellent position tracking performance of the designed control system.

抄録全体を表示

With the expectancy of higher performance of digital systems, the multiplier with both higher speed and higher precision is required as a key factor of the system improvement. On the other hand, it is quite common that these digital parts are assembled as a system LSI, and in this case, it is essential that these parts are re-usable as a core. The answer to this requirement is modularizing, which can cope with the variable precision, not a fixed one, as a multiplier structure.
The structure proposed here can perform a multiplication by cascading necessary number of the same modules and by repeated usage of the modules with necessary clock cycles. By this method, much higher speeding up compared to the conventional sequential type, as much as limitless multiplication precision compared to the parallel type, can be obtained.

抄録全体を表示

The diagnosis by qualitative reasoning is an effective way for complex systems such as a building air conditioning system. In reasoning process, the number of derived states exponentially increases with the size of the qualitative model because the states are derived based on all conceivable combinations of rule application.
This paper proposes a method of high-speed stochastic qualitative reasoning. In this method, model division and states composition are introduced. First, the partial models are constructed by dividing the entire model. Next, qualitative reasoning is executed in each partial model. Finally, the states in the entire model are generated by composing derived states. By this method, the number of states is reduced, and reasoning time is shorter than for one by the previous method and reasoning can finish regardless of model size.

抄録全体を表示

“Barrier free car” is specifically aimed at helping people who have trouble driving a normal-sized car due to age or disability. We propose a concept of a driving support system which compensates human error for providing safety drive.
Following research themes are discussed to provide a vehicle with computer aided driving support system.
1) Information re-composition
Grasp surrounding information of the vehicle and observe driver's behavior. Especially detection of pedestrians and recognition of face direction are described.
2) Intention recognition and Decision Action
Confirm safety depending upon following two-layered support to reduce the control by the system to a minimum.
•Predictive compensation; let a driver realize danger based on intention reasoning when miss-recognition or miss-decision happened.
•Fail safe intervention; control by support system at a safety limit.
3) Information Output
Provide drivers with the important information in the most suitable style in order not to disturb their task. Special features of our information output unit are the following four points.
Experimental and Reflective Cognition
Background and Foreground Mode
Media Transformation and Modality Transformation
Adaptable System Architecture

抄録全体を表示

The Internet, multimedia applications and digital devices have been ubiquitous in our society. Most people have their own PCs and can easily access to the World Wide Web site or on-line news. A system that can learn what information or programs a viewer wants or enjoys and that can then recommend them to the viewer would be a solution to the situations. In this paper, problems with conventional approaches are made clear by analyzing conventional methods of recommendation. Introducing reinforcement learning in society enables the system to recommend two kinds of serious programs. An evaluation of the new method is presented.

抄録全体を表示

We propose an algorithm to control the velocity of a mouse pointer so that it approaches to the intended object on a display by a user. And we show its usefulness through experiments. First, we describe following two facts through experiments;
a usual user moves a mouse pointer not drawing a strait-line but drawing an arc,
a beginner has trouble to point an object in its neighborhood.
Based on the above, we propose an algorithm to control the pointer as follows;
calculate the position of an imaginary object as weighted average of candidate objects,
control the velocity of the mouse pointer so that it approaches to the imaginary object.
And we show that the proposed algorithm improves mouse operation through experiments with human testers.

抄録全体を表示

Texts are frequently compressed for file size saving when they are stored in secondary storage. When we search compressed text files, it is usually necessary to decode them before search. Therefore, search is timeconsuming. On the other hand, we can use indexing for fast search. But this method does not consider compression ratio. We consider compression algorithms for file search systems with large amount of text files. Important factors of these methods are search time, expansion time and compression ratio. This paper proposes a two-stage compression method, which compresses text files using index files and compresses the result again with another algorithm. A preliminary experiment shows good performance in terms of com-pression ratio and search time when the same words appear frequently. We also discuss application of the two-stage compression method for large amount of text files such as newspaper and magazine data.

抄録全体を表示

Knowledge representation is important in the information processing. The local representation and the distributed representation are well known. Recently, area representation has attracted much attention because it can preserve advantages of each method. In this paper, we propose a novel neural network for four-term analogy based on area representation. It can deal with four-term analogy such as “teacher: student = doctor: ?” The proposed network is composed of three map layers and an input layer. The area representation method based on Kohonen Feature Map (KFM) is employed in order to represent knowledge, so that similar concepts are mapped in the nearer area in the map layer. The proposed mechanism in the map layer can realize the movement of the excited area to the near area. We carried out some computer simulations and confirmed the followings: (1) similar concepts are mapped in the nearer area in the map layer; (2) the excited area moves among similar concepts; (3) the proposed network realizes four-term analogy; (4) the network is robust for the lack of connections.

抄録全体を表示

The problem of robust tracking and model following is considered for a class of linear dynamical systems with multiple delayed state perturbations, time-varying uncertain parameters. and disturbance. A class of continuous memoryless state feedback controllers is proposed for robust tracking of dynamical signals. The proposed robust tracking controllers can guarantee that the tracking error decreases asymptotically to zero in the presence of multiple delayed state perturbations, time-varying uncertain parameters, and disturbance. A procedure for designing such a class of zero error tracking state feedback controllers is also introduced. Finally, a numerical example is given to demonstrate the validity of the results.

抄録全体を表示

This paper proposes a method creating a user profile for an information filtering system by genetic algorithm. Generally. user's interest is wide-ranged and unclear. In an information retrieval, it is difficult that a user expresses his query with suitable keywords. A proposed method extracts a user's perference from the documents which the user has read and evaluated. In addition, this method efficiently expores a search space for a user perference. This method uses simple genetic algorithm. We develop an information filtering system. and evaluate selection capability by this method. Finally, we compare this method with a relevance feedback in selection accuracy.

抄録全体を表示

In this paper, we propose an efficient procedure of physical parameter identification of a magnetic levitation system in closed-loop, where the levitated steel ball is controlled by a robust nonlinear controller which is designed based on rough nominal parameters.

抄録全体を表示

We propose a method for extracting QRS waves in an electrocardiogram (ECG) based on DP matching using an MIT/BIH arrhythmia database. The QRS wave is extracted to detect the characteristic points (CPs), the Q and S points, in an ECG. In the proposed method, a preprocessor, DP matching, and a detection algorithm are used. The preprocessor is used to divide an ECG into individual cardiac cycles. DP matching is used for selecting a template pattern. The detection algorithm is used to find CPs from the selected template pattern. The results of an experiment showed that SDs of detection errors using our method are 5.1 and 1.9 for the detection of the Q and S point, respectively. These small SDs indicate the reliability of our method.

抄録全体を表示

The ground-based wave distribution function method is effective and useful at any observable point on the ground to estimate the location and scale of the exit region of magnetospheric ELF/VLF radio waves on the ionospheric base. This is confirmed by numerical simulation, using the full-wave algorithm.

抄録全体を表示