G. Felici, F. Sun, and K. Truemper, "A Method for Controlling Errors in Two-Class Classification," Proc. 23rd Ann. Int'l Computer Software and Applications Conf. (COMPSAC-99), 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....large sets to ones of manageable size. The second rule eliminates common vectors of G(v) and B v w (v) since such vectors are not allowed in the subsequent computations. The third rule is another reduction step for large sets. Next, the learning step uses the data mining algorithm Lsquare of Felici, Sun, and Truemper (1999) to compute a set L v w (v) of logic formulas that correctly classify each g(v i ) resp. b v w (v j ) as being in G(v) resp. B v w (v) we cover this classification process later in this 6 section. Let us look ahead to the use of L v w (v) in the testing step. Suppose we have a testing text ....

....instead of just parts of speech, the use of a reference text for both learning and testing, and the way the characteristic vectors are evaluated. The next section covers the latter aspect. 19 6 Extraction of logic formulas Algorithm LEARN invokes the data mining algorithm Lsquare of Felici, Sun, and Truemper (1999) to extract logic formulas from sets of characteristic vectors. For completeness, we summarize the main features of Lsquare. Let two training sets A and B of f0; Sigma1g vectors of same length be given. Lsquare derives from A and B a collection L of 40 DNF logic formulas. When any one of the ....

Felici, G., Sun, F., and Truemper, K., A method for controlling errors in two-class classification, Proceedings of the 23rd Annual International Computer Software & Applications Conference COMPSAC 99, Phoenix, AZ, 1999, 186--191.

....thus providing also a compact explanation of the diagnosis. Lsquare has been developed in a recent collaboration between the Istituto di Analisi dei Sistemi ed Informatica del Consiglio Nazionale delle Ricerche (IASI CNR) and the University of Texas at Dallas and is described in detail in [7] and [8]. The paper is organized as follows. Sections 2 introduces some formal definitions of propositional logic. Section 3 summarizes the concepts of separating sets and the solution algorithm. In section 4 the Lsquare system is discussed. Section 5 describes the architecture of the database for ....

....We also have to state a non trivial conditions for separation to be possible. Consider two records of logic data, say g and f. We say that f is nested in g if for any entry f i of f equal to 1 or 1, the corresponding entry g i of g satisfies g i = f i . It can then be proved the following (see [8] for proof) Theorem: Let A and B be sets of 0, 1 records of the same length. Then a separating set S exists if and only if no record b B is nested in any record a A. From the above theorem a clear characterization of separating vectors can be given: a 0, 1 vector s separates of a ....

[Article contains additional citation context not shown here]

Felici G., Sun F-S., Truemper K. A Method for Controlling Errors in two-class classifications. XXIII Annual International Computer Software and Applications Conference. Phoenix, Arizona, October 2729, 1999

No context found.

G. Felici, F. Sun, and K. Truemper, "A Method for Controlling Errors in Two-Class Classification," Proc. 23rd Ann. Int'l Computer Software and Applications Conf. (COMPSAC-99), 1999.

No context found.

G. Felici, F. Sun, and K. Truemper, " A method for controlling errors in two-class classification", Proceedings of the 23 rd annual international computer software and applications conference (COMPSAC 99), Phoenix, AZ, 1999.

No context found.

G. Felici, F. Sun, and K. Truemper, " A method for controlling errors in two-class classification", Proceedings of the 23 rd annual international computer software and applications conference (COMPSAC 99), Phoenix, AZ, 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC