TY - GEN

T1 - On the sample complexity of consistent learning with one-sided error

AU - Takimoto, Eiji

AU - Maruoka, Akira

PY - 1993/1/1

Y1 - 1993/1/1

N2 - Although consistent learning is sufficient for PAC-learning, it has not been found what strategy makes learning more efficient, especially on the sample complexity, i.e., the number of examples required. For the first step towards this problem, only classes that have consistent learning algorithms with one-sided error are considered. A combinatorial quantity called maximal particle sets is introduced, and an upper bound of the sample complexity of consistent learning with one-sided error is obtained in terms of maximal particle sets. For the class of n-dimensional parallel axis rectangles, one of those classes that are consistently learnable with one-sided error, the cardinality of the maximal particle set is estimated and (Formula Found) upper bound of the learning algorithm for the class is obtained. This bound improves the bounds due to Blumer et al. [2] and meets the lower bound within a constant factor.

AB - Although consistent learning is sufficient for PAC-learning, it has not been found what strategy makes learning more efficient, especially on the sample complexity, i.e., the number of examples required. For the first step towards this problem, only classes that have consistent learning algorithms with one-sided error are considered. A combinatorial quantity called maximal particle sets is introduced, and an upper bound of the sample complexity of consistent learning with one-sided error is obtained in terms of maximal particle sets. For the class of n-dimensional parallel axis rectangles, one of those classes that are consistently learnable with one-sided error, the cardinality of the maximal particle set is estimated and (Formula Found) upper bound of the learning algorithm for the class is obtained. This bound improves the bounds due to Blumer et al. [2] and meets the lower bound within a constant factor.

UR - http://www.scopus.com/inward/record.url?scp=85029423545&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85029423545&partnerID=8YFLogxK

U2 - 10.1007/3-540-57370-4_53

DO - 10.1007/3-540-57370-4_53

M3 - Conference contribution

AN - SCOPUS:85029423545

SN - 9783540573708

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 265

EP - 278

BT - Algorithmic Learning Theory - 4th International Workshop, ALT 1993, Proceedings

A2 - Jantke, Klaus P.

A2 - Kobayashi, Shigenobu

A2 - Tomita, Etsuji

A2 - Yokomori, Takashi

PB - Springer Verlag

T2 - 4th Workshop on Algorithmic Learning Theory, ALT 1993

Y2 - 8 November 1993 through 10 November 1993

ER -