N. Ishii

Color coordination system on case based reasoning system using neural network

We propose a case-based reasoning (CBR) system whose case database consists of a neural network, and describe its application to a color coordination system. Furthermore, we apply a neural network to the CBR system to deal with fuzzy values which represent colors. The neural network in the CBR system, however, must learn a new case incrementally without forgetting the learned instances. To realize this ability, we use a new incremental learning method proposed by the us to reduce the computational complexity for the learning. In the new method, the system learns the generalized radial cases function both the new case and some old cases that are predicted and being interfered by the learning. For the experiments, we constructed a color coordination system for a make-up around eyes. The result gave appropriate combinations of colors that satisfied the user.

Sensory perception, learning and integration in neural networks

In biological visual neural networks, one of the prominent features is nonlinear functions, which play important roles in the visual system. However, the order of the nonlinearity of the visual system is one of the unsolved problems in its processing. The non-Fourier motion is visually perceived motion that cannot be explained simply by the autocorrelation (Fourier motion) of the stimulus. This non-Fourier motion is said to perceive it by the pre-processing of the nonlinearity transformation in the visual system. First, we analyze the structure and the function of the nonlinear asymmetric networks in the visual system. Second, sensory integration is realized by the sensor neural networks, which consist of the forward, backward networks and the integration neuron.

Combination of fast and slow learning neural networks for quick adaptation and pruning redundant cells

One advantage of the neural network approach is the learning of many instances with a small number of hidden units. However, the small size of neural networks usually necessitates many repeats of the gradient descent algorithm for the learning. To realize quick adaptation of the small size of neural networks, the paper presents a learning system consisting of several neural networks: a fast-learning network (F-Net), a slow-learning network (S-Net) and a main network (Main-Net). The F-Net learns new instances very quickly like k-nearest neighbors, while the S-Net learns the output of the F-Net with a small number of hidden units. The resultant parameter of the S-Net is moved to the Main-Net, which is only for recognition. During the learning of the S-Net, the system does not learn any new instances like the sleeping biological systems.

Structural analysis of steering wheel grip comfort by semantic differential method

Kansei quality varies with individual. Therefore, when building structural model of Kansei quality, it is very important to identify some essential structures by analyzing individual differences. In this work, semantic differential method is conducted in order to improve steering wheel grip comfort. Assuming a hierarchy structural model of the steering wheel grip comfort, the individual differences are analyzed using principal component analysis and then some hierarchy structural models are built using graphical modeling. As a result, two remarkable different structures are built. Some guidelines for developing steering wheel production can be shown by comparing the different structures.

Restoration of images via self-organizing feature map

S. Geman and D. Geman (1984) presented a basic statistical method for image restoration. In the method, the system searches an image X which makes a posterior probability p(X|Y) maximum, where Y is a noisy image given as an input. Using a Bayesian method, the posterior probability is rewritten as log p(X|Y)/spl prop/log p(X)+log p(Y|X), where p(X) and p(Y|X) are prior probability and likelihood of the image, respectively. The prior probability p(X) is usually represented by a heuristic function. S. Geman and D. Geman defined p(X) using the estimators to detect smoothness and edge. The prior probability greatly affects to the performance of the system so that it should be optimized to fit a class of images, which users want to restore. However, it is hard to optimize the estimator by hand. In this paper, we show a self-organizing feature map (SOM) proposed by Kohonen (1982) which approximately represents the prior probability of local features of images via learning. Therefore, the system can tune the estimator only by seeing original clean images. In the experiment section, we show that the new system using the SOM can restore actual images well.

On the Integration of SPC and APC: APC Can Be a Convenient Support forSPC

This paper is developed from Higashide et al. (Front Stat Qual Control 9:71–84, 2010). Automatic process control (APC) is frequently used in the semiconductor manufacturing process; however, statistical process control (SPC) is also needed to control the APC controller. This is an earlier paradigm on the integration of SPC and APC. Our viewpoint is different from the earlier one as follows: (a) APC reinforces SPC. (b) SPC complements APC.

Statistical Process Control for Semiconductor Manufacturing Processes

This paper considers statistical process control (SPC) for the semiconductor manufacuturing industry, where automatic process adjustment and process maintenance are widely used. However, SPC has been developed in parts industry an, thus, application of SPC to chemical processes such as those in the semiconductor manufacuturinghas not been systematically investigated

Structural analysis of steering wheel grip comfort by the semantic differential method

Abstract Kansei quality varies with the individual. Therefore, when building a structural model of Kansei quality, it is very important to identify some essential structures by analysing individual differences. In this paper, the semantic differential method is used to improve steering wheel grip comfort. Assuming a hierarchical structural model of the steering wheel grip comfort, the individual differences are analysed using principal component analysis, and then some hierarchical structural models are built using graphical modelling. As a result, two remarkably different structures are built. Some guidelines for developing steering wheel production can be shown by comparing the different structures.

Reconsidering Control Charts in Japan

In this paper, the role of control charts is investigated by considering the practice of using Shewhart control charts from the following viewpoints: 1) Which rules should be used for detecting an assignable cause? 2) What should be considered as chance cause? 3) What should be selected as control characteristics?

Supervised learning method for integrating information from several sensors-integration of inconsistent sensory inputs

In our previous work, we presented a sensory integrating system using several sets of neural networks and sensors. Each neural network recognizes inputs from corresponding sensors, the system integrates all outputs of the neural networks to get a high generalization ability. However, there are cases where the system fails to learn to discriminate some objects, if a part of the attributes of the object are shared by another class of objects. This is due to the fact that each neural network is independent from other networks. To solve the problem, we propose a new system which adaptively ignores a part of the sensory inputs which correspond to the shared attribute.