New equivalent circuits of neuron, different from the equivalent circuit of Hodgkin-Huxley, are proposed. In these circuits, ion-condensers are a concept to represent the equilibrium potential of ions, and power sources to represent ion-pumps. Uning these equivalent circuits, we derive the equations which relate the resting membrane potential with capacity of the ion-pumps, intracellular impermeant anions concentrations, extracellular ions concentrations and permeability of ions transported by the ion-pumps. Using these equations, we elaluate the effect of presence of intracellular impermeant anions on the resting membrane potential

View full abstract

A surgical operation is useful for the medical treatment of a brain tumor. The displacement of the tumor position due to change of the inner brain pressure from the position detected by MRI measurement is a great problem for the operation. It is almost impossible to detect the tumor visually due to the same color of the tumor tissue and the normal tissue. Sensing of magnetic field generated from the fine magnetic particles with a monoclonal antibody accumulated in the tumor tissue would be an effective way to detect the position of the tumor in the operation. A highly sensitive and potable micro magnetic sensor such as MI sensors having a resolution of 10^-5 Oe for DC field are required to detect such a very weak and localized field. Spatial distributions of a stray field of gel samples, in which fine magnetic particles are dispersed, were measured using the amorphous wire MI sensor aiming the brain tumor positioning. The spatial distributions were analyzed by an aligned dipole moment model. The calculation results using the model were good agreement with experimental results.

View full abstract

For producing of virtual reality contents based on real-time computer graphics, various techniques are studying actively. However studies whose object is the factors which are included in virtual reality contents and affect emotion of the observers, have not existed so far. On the other hand, at the studies called “Kansei Information Engineering”, several remarkable cases are reported, such as retrieving of photos or pictures and designing of dress or house by means of inputting “Kansei” words. For instance, they are “Warm”, “Modern”, “Romantic” and so on.
As a result that we paid attention to this means as an effective user interface, we developed a technique for adjustment of many kinds of parameters included in virtual reality contents. The parameters usually need to be adjusted by much trial and error labor. In our method, we have to make “Kansei” database which defines relationship between “Kansei” words and parameters. In order to make the database, first, we collected observers' subjective evaluation data by each “Kansei” word on several sample contents which are set on various parameters. Next, we analyzed the data by means of multiple regression analysis. Finally we defined the relationship as multiple regression formulas.

View full abstract

The problem of robust stabilization of a class of dynamical systems with partial knowledge of uncertainty bounds is considered. Here, the coefficient vector which fix the bounds of the uncertainties is assumed to be unknown. By combining a conventional robust state feedback controller with the proposed adaptive law which is used to update the bounds, we construct a new robust state feedback controller which can guarantee uniform ultimate boundedness of the resulting uncertain closed-loop systems. Furthermore, we apply it to river pollution control systems and show the validity of the adaptive robust controller proposed in this paper.

View full abstract

This paper studies how the generalization ability of modeling of the dynamic systems can be improved by taking advantages of second order derivatives of the criterion function with respect to the external inputs that can be considered as a direct implementation of the well-known regularization technique. Computation of second order derivatives of Universal Learning Network for a dynamic network are discussed. Simulation studies of a nonlinear dynamic system and a real robot system are carried out to demonstrate the effectiveness of the proposed method. Simulation results show that the proposed method can improve the generalization ability of neural networks sufficiently by selecting an appropriate regularization parameter.

View full abstract

A learning automaton is connected in feedback loop to the random environment, where the input to one side is the output to the other side. To use such abstract machine for formulating reinforcement learning systems, we need to generalize it into some new situations, which can correspond to context inputs from environment.
Here, we propose a new type learning automaton with context inputs on multi-teacher environments. For formulating some tradeoff environments, we present two descriptions, weighted majority decision and accumulated relative error decision.
We also present a strong theorem for proposed reinforcement scheme operating under some conditions. The simulation results show that, on static tradeoff environments, our reinforcement scheme has better properties in convergence speed and shaped learning processes.

View full abstract

To measure a distance and an inclination of a plane surface, various methods applying image processing have been developed in the fields of robot vision and automatic measuring systems of a three dimensional shape. The methods, however, need a complex optical system composed of light pattern sources and TV cameras, and need an image processing with complex algorithm. In addition, the measuring environment frequently needs to be well-conditioned to get good image data having a large SN ratio. From these viewpoints, the paper adopt to scan an optical range sensor to gather distance data between the sensor and various points on the plane surface. In applications, however, the distance data often include not only the sensor noise but also the setting errors accompanied with the sensor when the scanning data are gathered. The paper thus considers an optimal observation policy for measuring a distance and an inclination of a plane surface from the viewpoint of minimax criterion since a strictly optimal observation policy varies with those quantities of unknown plane surface.

View full abstract

This paper considers the robust stability of singularly perturbed systems with structured state space uncertainties. By making use of Lyapunov stability criterion and combining with the Lyapunov equations, a new approach for deciding a robust stability for uncertain linear singularly perturbed systems is presented. Based on the assumption that the reduced nominal system is stable, we also derive some sufficient conditions for robust stability. Some analytical methods and the Bellman-Gronwall inequality are used to investigate such sufficient conditions.
It is worth pointing out that in this paper, we do not need to investigate both the slow system and the fast system by using the singularly perturbation method because of the proposed method is very direct. Furthermore, we only assume that the uncertainties are a norm bounded. Therefore, the robust stability condition derived here is less conservative than those reported in the control literatures. An numerical example is given to demonstrate the validity of our new results.

View full abstract

This paper discusses a modeling of chaotic mobile robot avoiding moving obstacles under the orbit control with the intermittency chaos map. The study is aimed for the construction of the agent-oriented robot whose ultimate model is human thinking process and actions. The techniques proposed here were based on the new ideas of utilization of the intermittency chaos map, the arrangement of the sensor, and a parallel processing in obstacle avoidance and proceeding to a goal, with focus on the following points. (1) To avoid efficiently two or more moving obstacles by forecasting their movements. (2) To widen the range where a robot can move around so that it can move more freely and smoothly. (3) To try to make it manipulate itself with as intelligent actions seen in human as possible. In conclusion we obtained following results from the implementation of computer simulation above. (1) It is possible to get out of the area surrounded by two or more moving obstacles and reach the goal eventually for itself using the fluctuation. (2) The orbit control is possible which gives priority to safety though fairly careful movement is seen because of the fluctuation. (3) It is possible to give it more intellectual movement considering the priority of avoidance according to the movement of obstacles. Therefore, it follows from what has been said that the construction of the agent-oriented robot which could avoid the obstacles in a complex environment is possible by using the intermittency chaos map.

View full abstract

In this paper, we deal with the problem to find a static feedback gain matrix which satisfies multiple control specifications described in LMIs. This problem is easily solved when the intersection of all the sets of solutions of LMIs is not empty. However, even if it is empty, the gain matrix which satisfies all the specifications can exist. In this case, the problem to find such a matrix is equivalent to finding a linear subspace which intersects all the sets. We propose an algorithm to solve this problem. Some theoretical aspects are considered in a special case. Numerical experiments show the effectiveness of the algorithm.

View full abstract

High-speed transmission rates bring forward their specific issues influencing the network design. To cope with high-speed networks the routing algorithms should give a fast decision. The conventional routing algorithms of slow-speed networks are generally inapplicable for high-speed networks. The traffic control methods for high-speed networks must be adaptive, flexible, and intelligent for efficient network management. The intelligent routing algorithms based on Genetic Algorithms (GA), Neural Networks (NN), and Fuzzy Theory (FT) can prove to be efficient in high speed networks. In this paper, we propose a new adaptive routing method called ARGA (Adaptive Routing using GA). In the ARGA method, the network is modeled by a tree and the tree is reduced in the part where there are the same routes. In the reduced tree model, the tree junctions are represented by individual genes. By using a novel gene coding method, the chromosomes have the same length, which results in easy genetic operations and a fast routing can be achieved. Performance evaluation via simulations show that the ARGA method has a faster routing decision compared with the Genetic Load Balancing Routing (GLBR) method.

View full abstract

In recent years, we have been able to use very large scale databases through the Internet. However, user sometimes cannot retrieve information they need because of mismatching of words. Query expansion methods for information retrieval (IR) system have been studied. This paper proposes a new query expansion using fuzzy abductive inference for IR support system. The fuzzy abductive inference is the inference method based on Peng and Reggia's abductive inference. This method infers irredundant combinations of concept words for the query. Acquisition of relationships between concepts and terms is also studied in this paper.

View full abstract

In order to reduce the noise in the noisy speech, it is desirable to change the method of processing according to the intensity of the noise amount in order to reproduce a speech of good quality. Realizing this idea, we applied a three layered neural network which is able to learn the three graded speeches containing three graded white noise. Experimental results demonstrate that the noise amount could be estimated by the neural network, whose inputs are cepstral coefficients. We believe that the improvement of speech signals degraded by noise is fully accomplished by this system.

View full abstract

Redundant binary representations diminish the latency of addition at moderate hardware cost and thus can play an important role in processor design. Usually it had been overlooked that the choice of good representation for digits is of crucial importance to high speed and efficient implementation. Due to inadequate representation of digits, implementation of adders for redundant binary representations has been insufficient at speed.
In this paper we propose a high-performance redundant binary adder using a new hybrid representation method, which attempts to represent each digit by hybrid 2 bits/3 bits. This purpose is to present a compromise between the high speed of representing each digit by 3 bits and hardware-efficiency of representing each digit by 2 bits. By using the PARTHENON, a CAD system for VLSI, this adder is designed and evaluated. As a result, the proposed adder requires lower power dissipation, more efficient area, and about 1.4 times speed as compared with the conventional redundant binary adder. Finally, this proposed adder is applied to the multiplier and the availability is shown.

View full abstract