Catálogo de publicaciones - libros

This paper presents a new philosophy to model the behavior of a Student in a Tutorial- system using Learning Automata (LA). The model of the Student in our system is inferred using a higher-level of LA, referred to , which attempt to characterize the learning model of the Students (or Student Simulators), while the latter use the Tutorial- system. To our knowledge, this is the first published result which attempts to infer the learning model of an LA when it is treated externally as a black box, whose outputs are the only observable quantities. Additionally, our paper represents a new class of Multi-Automata systems, where the communicate with the Students, using LA, in a synchronous scheme of interconnecting automata.

In this paper, a novel approach Immune Algorithm (IA) is applied to improve the efficiency of composing near optimal test sheet from item banks to meet multiple assessment criteria. We compare the results of immune and Genetic Algorithm (GA) to compose test-sheets for multiple assessment criteria. From the experimental results, the IA approach is desirable in composing near optimal test-sheet from large item banks. And objective conceptual vector () and objective test-sheet test item numbers () can be effectually achieved. Hence it can support the needs of precisely evaluating student’s learning status. We successfully extend the applications of artificial intelligent - Immune to the educational measurement.

In view of the fact that most people do not know the names of the plants which can be seen everywhere, a plant search system with artificial intelligence on PDA devices is developed. In this study, users need to input the classified characteristics of leaves by observing plants’ leaves. After calculating the centroid-contour distance (CCD) and fuzzy function for all characteristics, the search results are displayed and arranged according to their ranks. Thus there is no need to consult all kinds of correlative books about plants. With the help of the plant search system on PDA devices, users can broaden their knowledge by mobile learning and the observation of plants.

This paper reports the first known solution to the Stochastic Point Location (SPL) problem when the Environment is non-stationary. The SPL problem [12,13,14] involves a general learning problem in which the learning mechanism attempts to learn a “parameter”, say , within a closed interval. However, unlike the earlier reported results, we consider the scenario when the learning is to be done . The Environment communicates with an intermediate entity (referred to as the Teacher) about the point itself, advising it where it should go. The mechanism searching for the point, in turn, receives responses from the Teacher, directing how should move. Therefore, the point itself, in the overall setting, is moving, delivering possibly incorrect information about its location to the Teacher. This, in turn, means that the “Environment” is itself non-stationary, implying that the advice of the Teacher is both uncertain and - rendering the problem extremely fascinating. The heart of the strategy we propose involves discretizing the space and performing a controlled random walk on this space. Apart from deriving some analytic results about our solution, we also report simulation results which demonstrate the power of the scheme.

In mining data streams, one of the most challenging tasks is adapting to , that is change over time of the underlying concept in the data. In this paper, we propose a novel ensemble framework for mining concept-changing data streams. This algorithm, called QACC (uick daptation to hanging oncepts), realizes quick adaptation to changing concepts using an ensemble of classifiers. For quick adaptation, QACC sensitively detects concept changes in noisy streaming data. Empirical studies show that the QACC algorithm is efficient for various concept changes.

Considering the interpretability of association classifier, and high classification accuracy of SVM, in this paper, we propose ACIK, an association classifier built with help of SVM, so that the classifier has an interpretable classification model, and has excellent classification accuracy. We also present a novel family of Boolean kernel, namely itemset kernel. ACIK, which takes SVM as learning engine, mines interesting association rules for construct itemset kernels, and then mines the classification weight of these rules from the classification hyperplane constructed by SVM. Experiment results on UCI dataset show that ACIK outperforms some state-of-art classifiers, such as CMAR, CPAR, L, DeEPs, linear SVM, and so on.

Social networks have been working as a medium to provide cooperative interactions between people. However, as some of users take malicious actions, the social network potentially contains some risks (e.g., information distortion). In this paper, we propose a robust information diffusion (or propagation) model to detect malicious peers on social network. Especially, we apply statistical sequence analysis to discover a peculiar patterns on recommendation flows. Through two experimentation, we evaluated the performance of risk discovery on social network.

In order to visualise the decision making process, the data association diagram is prepared to show the relationships or scenarios extracted from data, and provide a way for designing or discovering alternatives. However, managers are not easy to design alternatives if the collected data is large and complex. Thus, this study provides an approach for extracting the concepts from association diagrams to create the conceptual scenario diagrams. Afterward, variant diagrams are generated from the conceptual scenario diagrams for easily visualising and explaining the variation of financial status within a firm. Finally, the integrated scenario map is produced for managers to understand the financial manipulations of the firm.

Credit risk management based on portfolio theory becomes popular in recent Japanese financial industry. But consideration and modeling of chain reaction bankruptcy effect in credit portfolio analysis leave much room for improvement. That is mainly because method for grasping relations among companies with limited data is underdeveloped. In this article, chance discovery method with directed KeyGraph is applied to estimate industrial relations that are to include companies’ relations that transmit chain reaction of bankruptcy. The steps for the data analysis are introduced and result of example analysis with default data in Kyushu, Japan, 2005 is presented.

Automatic speech recognition for real-world applications such as a robot should deal with speech under noisy environments. This paper presents coarse phoneme recognition which uses a phoneme group instead of a phoneme as a unit of speech recognition for such a real-world application. In coarse phoneme recognition, the design of the phoneme group is crucial. We, thus, introduce two types of phoneme groups – exclusive and overlapping phoneme groups, and evaluate coarse phoneme recognition with these two phoneme grouping methods under various kinds of noise conditions. The experimental results show that our proposed overlapping phoneme grouping improves the correct phoneme inclusion rate by 20 points on average.