Catálogo de publicaciones - libros

We discuss the emerging applications of fuzzy logic and related technologies within the semantic web. Using fuzzy sets, we are able to provide an underlying semantics for linguistic concepts. We show how this framework allows for the representation of the types of imprecision characteristic of human conceptualization. We introduce some of the basic operations available for the representation and subsequent manipulation of knowledge. We illustrate the application of soft matching and searching technologies that exploit the underlying semantics provided by using fuzzy sets. We look at question-answering systems and point out how they differ from other information seeking applications, such as search engines, by requiring a deduction capability, an ability to answer questions by a synthesis of information residing in different parts of its knowledge base. This capability requires appropriate representation of various types of human knowledge, rules for locally manipulating this knowledge and framework for providing a global plan for appropriately mobilizing the information in the knowledge base to address the question posed. In this talk we suggest tools to provide these capabilities. We describe how the fuzzy set based theory of approximate reasoning can aid in the process of representing knowledge. We discuss how protoforms can be used to aid in deduction and local manipulation of knowledge. The concept of a knowledge tree is introduced to provide a global framework for mobilizing the knowledge in response to a query.

The present investigation is concerned with the crowd simulation in game or virtual reality and proposes new methodology dealing with emotion of the NPC (Non-Player Character) for increasing the reality of the behavior and action of crowd. The behavior of NPC depends on the individual disposition, which forms the properties of the crowd. The reorganization of the crowd is possible by meeting and parting according to the properties of NPC. In order to apply human emotion to the virtual characters, a number of factors and rules for identification of the status of emotion are considered. Fuzzy theory is used for the ambiguous description of the human emotion. The fuzzy functions and rules are designed to determine the conditions of emotion and reasonable inference is introduced to decide the control value of character’s actions like as speed and his direction. The proposed model is validated by the present experiments embodying more natural simulation of crowd behaviors.

The World Wide Web (Web) was not designed to ‘push’ information to clients but for clients to ‘pull’ information from servers (providers). This type of technology is not efficient in prompt information delivery from changing sources. Recently, XML-based ‘RSS’, or ‘Weblog’, has become popular, because they simulate real time information delivery using automated client pull technology. However, this is still inefficient because people have to manually manage large quantities of Web information, causing information overflow. Secondly, most current Web information still uses HTML instead of XML. Our automated information mediator (AIMS) collects new information from both traditional HTML sites and XML sites and alleviates the information overload problem by using narrowcasting from the server side, and information filtering from the client side using Multiple Classification Ripple-Down Rules (MCRDR) knowledge acquisition for document classification. The approach overcomes the traditional knowledge acquisition problem with an exception based knowledge representation and case based validation and verification. By employing this approach, the system allows domain experts, or even naive end users to manage their knowledge and personalize their agent system without help from a knowledge engineer.

Speech unit concatenation with a large database is presently the most popular method for speech synthesis. In this approach, the mismatches at the unit boundaries are unavoidable and become one of the reasons for quality degradation. This paper proposes an algorithm to reduce undesired discontinuities between the subsequent units. Optimal matching points are calculated in two steps. Firstly, the Kullback-Leibler distance measurement is utilized for the spectral matching, then the unit sliding and the overlap windowing are used for the waveform matching. The proposed algorithm is implemented for the corpus-based unit concatenating Korean text-to-speech system that has an automatically labeled database. Experimental results show that our algorithm is fairly better than the raw concatenation or the overlap smoothing method.

Usually feature extraction is applied for dimension reduction in hyperspectral data classification problems. Many researches show that nonparametric weighted feature extraction (NWFE) is a powerful tool for extracting hyperspectral image features and kernel-based methods are computationally efficient, robust and stable for pattern analysis. In this paper, a kernel-based NWFE is proposed and a real data experiment is conducted for evaluating its performance. The experimental result shows that the proposed method outperforms original NWFE when the size training samples is large enough.

This paper presents a new method that improves the performance of the speaker specific phonetic template based speech recognizer with the speaker-independent (SI) phoneme HMMs. Since the phonetic template based speech recognizer uses only the phoneme transcription of the input utterance, the performance of the system is worse than that of the speaker dependent system due to the mismatch between the training data and the SI models. In order to solve these problems, a new training method that iteratively estimates the phonetic templates and transformation vectors for the adaptation of the SI phoneme HMMs is presented. The phonetic class based and codebook-based stochastic matching methods are used to estimate the transformation vectors for speaker adaptation. Performance evaluation using the speaker dependent recognition experiments performed over actual telephone line showed a reduction of about 40% in the error rates when compare to the conventional speaker specific phonetic template based speech recognizer.

Ant Colony Optimization (ACO) is a collection of metaheuristics inspired by foraging in ant colonies, whose aim is to solve combinatorial optimization problems. We identify some principles behind the metaheuristics’ rules; and we show that ensuring their application, as a correction to a published algorithm for the vertex cover problem, leads to a statistically significant improvement in empirical results.

We model the well-known Sum-and-Product problem in a modal logic, and verify its solution in a model checker. The modal logic is public announcement logic. The riddle is then implemented and its solution verified in the epistemic model checker DEMO.

Plausible Logic is a non-monotonic logic with an efficient implementation. Plausible Logic has five proof algorithms, one is monotonic and four are non-monotonic. These five proof algorithms form a hierarchy. Ambiguity propagating proof algorithms are less risky than ambiguity blocking proof algorithms. The hierarchy shows that the benefit of using the riskier algorithms is that more formulas can be proved. Unlike previous Plausible Logics, the Plausible Logic in this paper is relatively consistent, checks for loops, can prove all its facts and all tautologies, and allows countably many formulas and rules to be considered.

We formulate the problem of transformation natural language sentences as the determination of sequence of actions that transforms an input sentence to its logical form. The model to determine a sequence of actions for a corresponding sentence is automatically estimated from a corpus of sentences and their logical forms with a MEM framework. Experimental results show that the MEM framework are suitable for the transformation problem and archived a comparable result in comparison with other methods.