Wyniki wyszukiwania - BazTech

1

Learning in Description Logics with Fuzzy Concrete Domains

Lisi F. A., Straccia U.

Fundamenta Informaticae

|

2015

|

Vol. 140, nr 3/4

373--391

EN

Description Logics (DLs) are a family of logic-based Knowledge Representation (KR) formalisms, which are particularly suitable for representing incomplete yet precise structured knowledge. Several fuzzy extensions of DLs have been proposed in the KR field in order to handle imprecise knowledge which is particularly pervading in those domains where entities could be better described in natural language. Among the many approaches to fuzzification in DLs, a simple yet interesting one involves the use of fuzzy concrete domains. In this paper, we present a method for learning within the KR framework of fuzzy DLs. The method induces fuzzy DL inclusion axioms from any crisp DL knowledge base. Notably, the induced axioms may contain fuzzy concepts automatically generated from numerical concrete domains during the learning process. We discuss the results obtained on a popular learning problem in comparison with state-of-the-art DL learning algorithms, and on a test bed in order to evaluate the classification performance.

2

Bisimulation-Based Concept Learning in Description Logics

Tran T.-L., Ha Q.-T., Hoang T.-L.-G., Nguyen L. A., Nguyen H. S.

Fundamenta Informaticae

|

2014

|

Vol. 133, nr 2/3

287--303

EN

Concept learning in description logics (DLs) is similar to binary classification in traditional machine learning. The difference is that in DLs objects are described not only by attributes but also by binary relationships between objects. In this paper, we develop the first bisimulation-based method of concept learning in DLs for the following setting: given a knowledge base KB in a DL, a set of objects standing for positive examples and a set of objects standing for negative examples, learn a concept C in that DL such that the positive examples are instances of C w.r.t. KB, while the negative examples are not instances of C w.r.t. KB. We also prove soundness of our method and investigate its C-learnability.

3

A Dichotomic Search Algorithm for Mining and Learning in Domain-Specific Logics

Ferré S., King R. D.

Fundamenta Informaticae

|

2005

|

Vol. 66, nr 1,2

1--32

EN

Many application domains make use of specific data structures such as sequences and graphs to represent knowledge. These data structures are ill-fitted to the standard representations used in machine learning and data-mining algorithms: propositional representations are not expressive enough, and first order ones are not efficient enough. In order to efficiently represent and reason on these data structures, and the complex patterns that are related to them, we use domain-specific logics. We show these logics can be built by the composition of logical components that model elementary data structures. The standard strategies of top-down and bottom-up search are ill-suited to some of these logics, and lack flexibility. We therefore introduce a dichotomic search strategy, that is analogous to a dichotomic search in an ordered array. We prove this provides more flexibility in the search, while retaining completeness and non-redundancy. We present a novel algorithm for learning using domain specific logics and dichotomic search, and analyse its complexity. We also describe two applications which illustrates the search for motifs in sequences; where these motifs have arbitrary length and length-constrained gaps. In the first application sequences represent the trains of the East-West challenge; in the second application they represent the secondary structure of Yeast proteins for the discrimination of their biological functions.

4

Learning classification programs : the Genetic Algorithm approach

Giordana A., Lo Bello G.

Fundamenta Informaticae

|

1998

|

Vol. 35, nr 1-4

163-177

EN

Genetic Algorithms have been proposed by many authors for Machine Learning tasks. In fact, they are appealing for several different reasons, such as the flexibility, the great exploration power, and the possibility of exploiting parallel processing. Nevertheless, it is still controversial whether the genetic approach can really provide effective solutions to learning tasks, in comparison to other algorithms based on classical search strategies. In this paper we try to clarify this point and we overview the work done with respect to the task of learning classification programs from examples. The state of the art. Emerging from our analysis suggests that the genetic approach can be a valuable alternative to classical approaches, even if further investigation is necessary in order to come to a final conclusion.