This paper describes some properties of storage capacity and robustness of correlation associative memory of sequential patterns using higher order neural networks. First, it is shown that storage capacities for k = 1, 2 and 3 dimensional cases are 0.263N, 0.227(^N₂) and 0.197(^N₃) from the prediction using the transition properties, respectively, where N is the number of neurons and (^N_k) means the combination of k from N. And it is shown that higher order models are superior in the pattern selection ability to the conventional ones. Further, it is shown that higher order correlation models have high robustness compared to the conventional ones.

抄録全体を表示

In this paper, a new approximate stability analysis method for discrete-time basis nonlinear dynamic systems using higher order derivatives of Universal Learning Networks(ULNs) is proposed. The proposed method is essentially the same as the coventional ones that linearize the nonlinear systems and then use the linear stability analysis, but the stability of the system is analyzed by calculating the forward propagation algorithm of ULNs in the proposed method. Furthermore, the domain of being able to linearlize the nonlinear dynamical systems is obtained by finding the area where the first order derivatives are dominant, which is hard to find in the conventional methods. Therefore, whether the dynamical system in this domain is stable or not is studied. Simulations of an inverted pendulum balancing system are carried out to study the effectiveness of the proposed method. From the results of the simulations, it is clarified that the stability of the inverted pendulum control system is easily analyzed by the proposed method.

抄録全体を表示

In this paper, we propose a method to analyze the vibration influence for a wrist by Motor Operated Electric Tools (MOETs) using the Self-Organizing features Map (SOM). We measure twenty-seven data to evaluate the vibration in the wrist for three kinds of MOETs. In recent years, the International Organization for Standardization (ISO) has suggested the sensitivity values of vibration for wrists. We develop a new classification method using the spectrum being adjusted by this sensitivity effects based on the SOM. The classification method shows the effectiveness for classification of various types of MOETs, which are suitable for operators from viewpoints of human effects of the vibrations.

抄録全体を表示