Sosuke Iwai

An intelligent data-base interface using psychological similarity between data

A question-answering system for data-base interfaces is presented which uses psychological similarity between data. The similarity relationships between data are derived from a data base that is based on a set-theoretic model of psychological similarity. These relationships are represented in the computer as a network. The generalization function enables propagation of information obtained from user similarity responses over the network. Using the generalization function, the computer determines the generic kind of question it should next pose to the user. Through this question-answering process, the knowledge-based system aids the user in specifying requests for relevant data, as well as in retrieving data from the data base. Finally, the system presented here is applied to a chemical data base, and the results of the question-answering process with the implemented system are discussed.

Fusing multiple data and knowledge sources for signal understanding by genetic algorithm

This paper presents a new approach to partially automating a human experts proficient interpretation skills for data and knowledge fusion in signal-understanding tasks. The authors start by recognizing the fact that signal interpretation is attributed much to a human experts domain-specific, pattern-perceiving capability of grasping raw signals by structured representations having multiple levels of abstraction, rather than to some objectively defined knowledge. In other words, that is an emergent or self-organizing process, where information is regarded as perceptual as opposed to objectively defined. First, they attempt to organize such structured representations by usage of a hierarchical clustering method of data analysis. Then, based on these representations they model a human experts interpretation skill as an activity of searching for an optimum combination of those perceptual units within that structured representation space being constrained by the data. In order to implement this activity, they introduce a genetic algorithm and apply it to the structured representation space assimilating a human analysts creative interpreting task in flexibly shifting the focal view of attention from the coarse to the precise. They implement a working system for signal understanding of the remote sensing data of seismic prospecting and show the results output by the system.

AN ADAPTIVE CNC SYSTEM OF A MILLING MACHINE TOOL

An adaptive CNC (computer numerical control) system of a milling machine tool using a precise tool wear rate model is presented. In the system, the bending moments on the spindle are sensed and used for computing the fundamental parameters of cutting conditions such as work material hardness, cutter sharpness, and others. The flank wear land temperature of the tool tip is computed from these parameters. Then, the tool wear rate is expressed as a function of these parameters and the flank wear land temperature, which is essential information to compute the performance index representing productivity or efficiency of the system. Results of preliminary off line cutting experiments verify that the above mentioned methods to compute the parameters, the temperature, and the tool wear rate are correct. An adaptive control cutting test adjusting the feedrate to keep the tool wear rate at constant value shows that this system can successfully reply to the variations of the tool wear phenomena. A method to identify the other important cutting condition parameters such as width and depth of cut is also studied.

Integrated cognitive architecture for image understanding using fuzzy clustering and structured neural network

The authors present an architecture for an integrated image understanding system, in which the following two aspects of human uncertainty-handling in visual information processing are modeled: feature extraction from fuzzy input images, and consistent interpretation by dynamically reducing uncertainties contained in the images. An integrated cognitive architecture is proposed, in which both aspects are modeled by using fuzzy clustering methods and a structured neural network, respectively, which are jointed and coupled to realize the cooperative understanding of an image. The system was implemented for understanding geological images obtained from remote-sensing satellites.<<ETX>>

Chaotic behavior of fuzzy symbolic dynamics and its relation to constraint-oriented problem solving with hierarchical structures reflecting natural life

By introducing fuzziness as coding schema of continuous variables into discrete and symbolic values, several methods of constraint-oriented problem solving are discussed. The hierarchical nature of these methods are examined and are shown to be tightly related to the hierarchical structure of natural systems, particularly, of biological systems. Also, two different kinds of computational paradigms in these hierarchies are introduced and are shown to be able to yield chaotic behavior due to the fluctuations, caused by the above fuzziness-based coding from continuous variables into symbolic values. Moreover, the importance at the effect of the chaotic behavior on constraint-oriented problem solving is also elucidated.<<ETX>>

Decision-theoretic selection of reasoning scheme for an autonomous robot under resource constraints

Investigates a formally rigorous technique based on decision-theoretic principles that address the problem of bounding the amount of reasoning cost under resource constraints and uncertainty. The authors derive a set of optimal decision policies that may be used to decide which reasoning scheme to use and where it should be used. The decision model, which is represented by an influence diagram, takes into account all sources of uncertainty and risk preferences in order to derive the optimal decision policy by deliberating on the trade-off between the value of the information that is potentially available and the computational cost. By developing such a metalevel decision model, the authors construct an intelligent agent that is embedded in the environment and can behave in a sophisticated fashion with minimal environmental interactions. The authors apply such an agent architecture to an autonomous mobile robot with limited sensing and actuating capabilities, and present the simulated behaviors which are supervised by their proposed model.

A knowledge acquisition system for conceptual design based on functional and rational explanations of designed objects

Abstract An EBL-based system for acquiring conceptual design knowledge in physical systems was proposed and implemented based on Value Engineering Methodologies and Axiomatic Design Approaches. The key idea behind this system is that such knowledge can be acquired by analyzing existing designed objects which can be regarded as the result of rational decisions and actions. In this system, the structural features of designed objects are analyzed by domain specific knowledge to yield a systematic explanation of how they function and attain their design goals and why they are used for attaining the goals. The “how” explanation results in a generalized version of the Functional Diagram used in Value Engineering from which the object in question can be interpreted via two kinds of design rationalities, i.e., teleological and causal ones. Namely, a designed object is regarded as being consisted of a hierarchy of design goals (primary functions), subgoals (subfunctions), structures and substructures toward attaining those goals. We applied the EBL method to this hierarchical model to acquire general design knowledge which is operational at various phases and fields in the conceptual design. Through organizing domain-specific knowledge of this EBL system according to the above hierarchical model, various levels of knowledge can be extracted by a single positive instance. The quality of the extracted knowledge is then discussed with reference to its level in the hierarchy of acquisition. The “why” explanation gives us a deeper understanding of the designed objects from which we can then extract meta-planning or strategic knowledge for selecting rational plans from among other possible alternatives. This deep explanation is obtained by regarding the object in question as being the result of a sequence of strategically rational decisions and actions which are subject to the principles of “good” design that are formalized as an axiomatic system in the Axiomatic Design Approach.

Real time programming of computer numerical control of a machine tool in CAM

Abstract A method of real time programming of computer numerical control which composes a small distributed on-line control system in CAM is presented. A simple multiprogramming method in which DDC, numerical control and adaptive control are simultaneously performed by using only a level of interruption with a clock is presented. Methods to decrease the computing time of DDC and interpolation in numerical control by using single precision calculations are presented. It has been verified that the path error and the time occupation rate of the presented interpolation method are small enough. Programming of adaptive control which identifies the parameters of a cutting process which are necessary for the automatic determination of the optimum cutting speed and feedrate is also presented. It has been verified by experiments that the adaptive control and other control programs function well with no trouble about the CPU time by using the presented methods.

A Human-Friendly Interface System for Decision Support Based on Self-Organized Multi-Layered Knowledge Structures

An architecture for decision support system based on the knowledge represented as causally-chained networks in socio-political domain is presented. Human experts’ decisionmaking is characterized by their simplifying and pattern-perceiving heuristics, which enables them to make sense however complex reality they may face to. The system is designed to realize capabilities to generate dynamic evolving patterns of events with different abstraction levels in a hierarchical multi-layered fashion above the empirical knowledge store, capturing event concepts using a meta-knowledge universally accepted as human social behaviours. Based on such a human memory-like structure, an interface system is implemented enabling an easy and flexible access to the knowledge base supporting decisionmakers’ memory and conceptualization phases in explicating the problem in their pre-decision stages.

Signal understanding of spectrum data using Bayesian network and neural network

This paper presents a method for automating the task of understanding signals of spectrum data. Probabilistic inferences are used for the purpose of decision theoretic fusion of multi-source data and knowledge, and a neural network is implemented within that as a subprocess that provides with pattern-specific concepts to the fusion model. The difficulties in using neural networks are: 1) vague transparency of the learned concepts; and 2) a screening problem of the training data to attain plausible learning. By restricting the concepts to be learned by the neural network only to the pattern-specific concepts and by joining it with another transparent probabilistic reasoning scheme, our proposing architecture could overcome the above problems.