Ebook: Advances in Logic Based Intelligent Systems

It constitutes a great honor for us to publish the selected papers of the 5th Congress of Logic Applied to Technology – LAPTEC'2005 held in Himeji, JAPAN, from April 2nd to 4th, 2005. LAPTEC'2005 was hosted School of Human Science and Environment – University of Hyogo in Japan. It was the first time for LAPTEC to be held in other countries than Brazil since its birth in 2000, and has made the congress more international, with delegates from Japan, Brazil, Taiwan, China, Australia, Brunei. In LAPTEC'2005 we promoted discussion and interaction between researchers and practitionaers focused on both theoretical and practical disciplines concerning logics applied to technology, with diverse backgrounds including all kinds of intelligent systems having classical or non-classical logics as underlying common matters.

First of all, we would like to express our greatest gratitude to Dr. Isao Shirakawa who accepted our offer to be the general chair of LAPTEC'2005, and Dr.Yutaka Suzuki (Univ. Hyogo, Vice President) and Dr.Akihiro Amano (Univ. Hyogo, Vice President) who have kindly presented their great invited lectures in the congress. We also would like to express our gratitude to Dr. Tetsuya Murai and Dr. Masahiro Inuikuchi who have organized the workshop “Rough Sets and Granularity” in LAPTEC'2005, Prof. Germano Lambert Torres (Univ. Itajuba - Brazil) and his staff for the construction of the web site of the LAPTEC2005 and its maintenance, and all other committee members.

We have chairs and committees in LAPTEC'2005 as

General Chairs: Isao Shirakawa (Japan), Kazumi Nakamatsu (Japan), Jair Minoro Abe (Brazil)

Honorary Committee: Hiroakira Ono (Japan), Kiyoshi Iseki (Japan), Lotfi A. Zadeh (U.S.A.), Newton C.A. da Costa (Brazil), Patrick Suppes (U.S.A.), Yutaka Suzuki (Japan)

Program Committee: Ajith Abraham (U.S.A.), Atsuyuki Suzuki (Japan), Don Pigozzi (U.S.A.), Edger G. K. Lopez-Escobar (U.S.A.), Eduardo Massas (Brazil), Germano Lambert Torres (Brazil), Hiroakira Ono (Japan), John A. Meech (Canada), Lakhmi C. Jain (Australia), Lotfi A. Zadeh (U.S.A.), Kenzo Kurihara (Japan), Kiyoshi Iseki (Japan), Manfred Droste (Germany), Marcelo Finger (Brazil), Maria C. Monard (Brazil), Michiro Kondo (Japan), MuDer Jeng (Taiwan), Nelson Favilla Ebecken (Brazil), Newton C.A. da Costa (Brazil), Patrick Suppes (U.S.A.), Pulo Veloso (Brazil), Sachio Hirokawa (Japan), Seiki Akama (Japan), Setsuo Arikawa (Japan), Sheila Veloso (Brazil), Sheng-Luen Chung (Taiwan), Shusaku Tsumoto (Japan), Tadashi Shibata (Japan), Takahira Yamaguchi (Japan), Tetsuya Murai (Japan), Yukihiro Itoh (Japan), Yutaka Hata (Japan).

Organizing Committee: Alexandre Scalzitti (Germany), Claudio Rodrigo Torres (Brazil), Hiroshi Ninomiya (Japan), Kazuo Ichikawa (Japan), Marcos Roberto Bombacini (Brazil), Patrick T. Dougherty (Japan), Yutaka Yamamoto (Japan).

Also we would like to thank the following scholars who helped us in refereeing papers: Maria Ines Castineira (Brazil), Claudia Regina Milare (Brazil), Ricardo Luis de Freitas (Brazil), Adenilso da Silva Simao (Brazil).

Last, we would like to express great thanks to University of Hyogo with hosting LAPTEC'2005, and acknowledge that this publication was partly supported by the Grant in The Japanese Scientific Research Fund Foundation (C)(2) Project No. 16560468.

This book is dedicated to Emeritus Professor Atsuyuki Suzuki in commemoration of his honorable retirement from Shizuoka University, March 2005. Prof. Suzuki is learned in application of paraconsistent logic, and has contributed many papers and as a member of the program committee to LAPTEC since the beginning.

Kazumi Nakamatsu, Jair Minoro Abe, Chairs LAPTEC'2005

Infon Logic was introduced by Devlin as a logic for situation theory which constitutes a logical foundation for Barwise and Perry's situation semantics. In this paper, we propose constructive infon logic (CIL) based on Nelson's constructive logic with strong negation. A Kripke semantics is given with a completeness proof.

This paper presents a Hybrid Particle Swarm Optimizers combining the idea of the particle swarm with concepts from Evolutionary Algorithms. The hybrid Particle Swarm Optimizers with Mutation (HPSOM) combine the traditional velocity and position update rules with the idea of numerical mutation. This model is tested and compared with the standard PSO on unimodal and multimodal functions. This is done to illustrate that PSOs with mutation operation have the potential to achieve faster convergence and the potential to find a better solution. The objective of this paper is to describe the HPSOM model and to test their potential and competetiveness on function optimization.

Simplification orderings, like the recursive path ordering and the improved recursive decomposition ordering, are widely used for proving the termination property of term rewriting systems. The improved recursive decomposition ordering is known as the most powerful simplification ordering.

In this paper, we investigate the improved recursive decomposition ordering for proving termination of term rewriting systems. We completely show that the improved recursive decomposition ordering is closed under substitutions.

Data transformation is a kind of data preprocessing [1, 3, 5] and an important procedure for mathematical modelling. Mathematical model estimated based on a training data set results better if the data set has been properly preprocessed before passed to the modelling procedure. In the paper, different preprocessing methods on automotive engine data are examined. The preprocessed data sets using different preprocessing methods are passed to neural networks for models estimation. The generalizations of these estimated models could be verified by applying test sets, which determine the effects of different preprocessing methods. The results of preprocessing methods for automotive engine data are shown in the paper.

The Fully Bayesian Significance Test (FBST) is a coherent Bayesian significance test for sharp hypotheses. This paper explores the FBST as a model selection tool for general mixture models, and gives some computational experiments for Multinomial-Dirichlet-Normal-Wishart models.

This article presents an hybrid computational model , called Neo-Fuzzy-Neuron Modified by Kohonen Network (NFN-MK),that combines fuzzy system techniques and artificial neural networks. Its main task consists in the automatic generation of membership functions, in particular, triangle forms, aiming a dynamic modeling of a system. The model is tested by simulating real systems, here represented by a nonlinear mathematical function. Comparison with the results obtained by traditional neural networks, and correlated studies of neurofuzzy systems applied in system identification area, shows that the NFN-MK has a similar performance, despite its greater simplicity.

Process control in a brewery plant deals with the open/close decisions of valves for pipelines in the brewery to meet the service requests of filtration and CIP (Clean in pipe) processes. In order to maximize concurrency among different process requests, it is desired that non-conflicting processes be enabled as much as possible. By exploring its similarity to that of railway interlocking policy, this paper adopts an EVALPSN-based concurrency control approach proposed by Nakamatsu et al. In doing so, system configuration of the system in terms of sub-process and valves for all the processes involved is first tabulated. EVALPSN-statements that reflect the pipeline configuration and imposed safety constraints of mutual exclusive usage of sub-processes are then systematically constructed. In deriving a decision as to either granting or denying a service request, these EVALPSN statements are executed in a PLC-based implementation that is connected to both human operator's input requests as well as sensor status updates. Successfully implemented for a local brewery plant in Taiwan, the EVALPSN-based decision approach is shown to have the advantage in general pipeline control applications.

This work shows a process of decision making based on a new kind of logic - Paraconsistent Annotated Logic (PAL). Choosing the factors that influence in the success or in the failure of an enterprise and applying the PAL techniques, the Para-analyser Algorithm and the Baricenter Analysis Method we obtain a sole result. Then we can decide if the enterprise is viable or not viable, or if the data are non-conclusive, with an established level of requirement.

We have developed an annotated logic program called an Extended Vector Annotated Logic Program with Strong Negation(abbr. EVALPSN), which can deal with defeasible deontic reasoning and contradiction. We have already applied EVALPSN to safety verification and control such as railway interlocking safety verification. In this paper, we show pipeline valve safety verification to avoid liquid mixture accidents with a simple example for brewery pipeline control.

In this paper, we introduce a typical discrete event control example, Cat and Mouse problem can be controlled by EVALPSN stable model computation. First we show the Cat and Mouse example can be easily formalized as an EVALPSN whose stable models provide its control. Generally, stable model computation takes long time and not so appropriate for real-time control. Therefore, in order to realize real-time control for the Cat and Mouse example, we consider a restricted subset of the stable models.

Driving actions of human beings such as putting the brake in order to control the car speed can be regarded to be decided by defeasible deontic reasoning based on environmental information such as the distance between two cars. We formalize such a car driving model in a paraconsistent logic program EVALP(Extended Vector Annotated Logic Program), which can deal with defeasible deontic reasoning. In this paper, we introduce an EVALP defeasible deontic reasoning based car driving model and a traffic simulation system based on the model, which can be implemented in the cell automaton method for traffic simulation.

Modern automotive engines are controlled by the electronic control unit (ECU). The electronically controlled automotive engine power & torque is significantly affected with effective tune-up of ECU. Current practice of ECU tune-up relies on the experience of the automotive engineer. Therefore, engine tine-up is usually done by trial-and-error method because a mathematical power & torque model of the electronically controlled engine has not been determined yet. With an emerging technique, Support Vector Machines (SVM), the approximate power & torque model of an electronically controlled vehicle engine can be determined by training the sample data acquired from the dynamometer. This model can be used for the engine performance prediction. The construction and accuracy of the model are also discussed in this paper. The study shows that the predicted results are good agreement with the actual test results.

The supervised machine learning approach usually requires a large number of labelled examples to learn accurately. However, labelling can be a costly and time consuming process, especially when manually performed. In contrast, unlabelled examples are usually inexpensive and easy to obtain. This is the case for text classification tasks involving on-line data sources, such as web pages, email and scientific papers. Semi-supervised learning, a relatively new area in machine learning, represents a blend of supervised and unsupervised learning, and has the potential of reducing the need of expensive labelled data whenever only a small set of labelled examples is available. Multi-view semi-supervised learning requires a partitioned description of each example into at least two distinct views. In this work, we propose a simple approach for textual documents pre-processing in order to easily construct the two different views required by any multi-view learning algorithm. Experimental results related to text classification are described, suggesting that our proposal to construct the views performs well in practice.

In the development of complex distributed systems using a cognitive multi-agent approach, where each agent encapsulates an expert system, the knowledge acquisition process is known to be the most difficult task. This paper presents a methodology based on planning and on high level Petri nets to the knowledge acquisition process of such systems. This methodology was applied in the implementation of a robot soccer team for the Robocup simulator.

This paper is a survey which presents fundamental ideas about the application of weighted automata in digital image processing. We present basic definitions such as semirings, weighted automata and digital images. Then we explain how we can represent an image using a weighted automaton.

It can be observed that the number and the complexity of the application domains, where the Paraconsistent Annotated Logic has been used, have grown a lot in the last decade. This increase in the complexity of the application domain is an extra challenge for the designers of such systems, once there are not suitable computerized models for the representation and abstraction of the paraconsistent systems. This work proposes a new model to Paraconsistent Systems called Paraconsistent Finite Automata.

In this paper we presented a System capable to realize a recognition of characters with base in the theoretical concepts of the Paraconsistent Annotated Logic. The Paraconsistent Annotated Logic PAL as shown in [1] is a class of the Non-Classic Logic which allows to manipulate contradictory signals. In [5] were presented the Paraconsistent Artificial Neural Cells built with Algorithms based on PAL. These Cells showed the capacity of learning certain signals in form of functions applied in their inputs. In this work, based on these Cells, were made connections and groupings among the algorithms to create a Recognizer of Characters Paraconsistent System (RCPS) capable of to learning and recognizing different types of alphabet letters or sources of signals. After the learning characters, the RPCS can recognize the letter with a high efficiency and further compares it to the group of characters learned previously. The results of tests demonstrate that the RPCS can be used as Specialist Systems of words and images Recognition

Feature Subsect Selection is an important issue in machine learning, since non-representative features may reduce accuracy and comprehensibility of hypotheses induced by supervised learning algorithms. Feature Subsect Selection is applied as data pre-processing step, which aims to find a subset of features that describes well the data to be used as input to the inducer. Several approaches to this problem have been proposed, among them the filter approach. This work proposes a filter which uses Fractal Dimension as importance criterion to select a subset of features from the original data. Empirical results on real world data sets are presented. Performance comparison of the proposed criterion with two other criteria frequently considered within the filter approach shows that Fractal Dimension is an appropriated criteria to select features for supervised learning.

In digital electronics there may be situations in which the designer may need the system to be able to “undo” certain processes. If that is a requirement, what the designer does, in simple mathematical terms, is to restrict the domain so that the process becomes a one-to-one function. The purpose of this short note is to discuss some simple logical operations in computer science in the context mathematical functions.

In this paper, we present a new learning algorithm of neural network based on the orthogonal decomposition method (ORT). The main scheme of this algorithm is using the ORT to obtain specially structured subspaces defined by the input-output data. This structure is then exploited in the calculation of the parameter estimation of the neural network. Therefore, the method to obtain the comparatively accurate estimate is introduced without iteration calculations. We show that this algorithm can be applied to successfully identify the nonlinear system in the presence of comparatively loud noise. Results from several simulation studies have been included to the effectiveness of this method.

In this expository work we show how the Para-analyzer can be useful to a variety of applications involving decision-making when facing mainly with uncertainty, inconsistent or paracomplete information. The Para-analyzer can be implemented electronically, originating the Para-control, very useful in applications in the area of Robotics and Automation.

Practical Data Mining applications use learning algorithms to induce knowledge. Thus, these algorithms should be able to operate in massive datasets. Techniques such as dataset sampling can be used to scale up learning algorithms to large datasets. A general approach associated with sampling is the construction of ensembles of classifiers, which can be more accurate than the individual classifiers. However, ensembles often lack the facility to explain its decisions. In this work we explore a method for constructing ensembles of symbolic classifiers, such that the ensembles are able to explain its decisions to the user. This idea has been implemented in the ELE system described in this work.

In this work we present an approach to extract and to structure bibliographical references from BibTex files, allowing the identification of the duplicate ones, which can appear slightly different in different files. To deal with this problem, existing systems use classifiers, clustering or others algorithms, allied with an Edit Distance metric, to distinguish between duplicate and nonduplicate records. The main challenge is to identify the duplicate records in database where the volume of the references can reach millions, in an efficient computational time. The technique proposed constructs a key (string) with information from each reference and stores them in a metric data structure called Slim-Tree. The Slim-Tree structure allows the minimization of the comparisons between references (being close to O(n log (n))), considering only the most similar keys to a given one.

We clarify the relation between autoepistemic theories and a paraconsistent logic program called Vector Annotated Logic Program (VALPSN for short) proposed by K.Nakamatsu et al. We review the stable class semantics for VALPSN and propose a translation from Moor's autoepistemic theories to VALPSN. Based on the translation, we prove that there is a one-to-one correspondence between stable classes of VALPSN and iterative expansion classes of autoepistemic theories.