Relational concepts and the fourier transform: An empirical study
Entity
UAM. Departamento de Ingeniería InformáticaPublisher
Springer Berlin HeidelbergDate
1998Citation
10.1007/3-540-64575-6_67
Advances in Artificial Intelligence: 12th Biennial Conference of the Canadian Society for Computational Studies of Intelligence, AI'98 Vancouver, BC, Canada, June 18–20, 1998. Proceedings. Lecture Notes in Computer Science, Volumen 1418. Springer, 1998. 411-425
ISSN
0302-9743 (print); 1611-3349 (online)ISBN
978-3-540-64575-7 (print); 978-3-540-69349-9 (online)DOI
10.1007/3-540-64575-6_67Funded by
This research was supported in part by grant IRI 92-04473 from the US National Science Foundation. We thank Lenny Pitt for bringing to our attention the literature on learning through the Fourier transform, and for his suggestions.Editor's Version
http://dx.doi.org/10.1007/3-540-64575-6_67Subjects
Artificial Intelligence; InformáticaNote
The final publication is available at Springer via http://dx.doi.org/10.1007/3-540-64575-6_67Proceedings of 12th Biennial Conference of the Canadian Society for Computational Studies of Intelligence, AI'98 Vancouver, BC, Canada, June 18–20, 1998
Rights
© Springer-Verlag Berlin Heidelberg 1998Abstract
Lack of domain knowledge may impose primitive data representations. Then, complex (non-linear) relationships among attributes complicate learning, especially for typical learning methods. These methods fail because their bias does not match the complex relational structures relevant to the domain. However, more recent approaches to learning have implemented biases that allow learning of structured, albeit complex, concepts. One of such approaches, based on the Fourier transform of Boolean functions, is studied and compared empirically to others, based on constructing new features or extracting relations from propositional training data. Controlled experiments help to characterized the kinds of concept that allow each approach to outperform the others. This characterization, which implicates parameters of Fourier complexity, other measures of concept difficulty, and the relational structure of the target concepts, is also discussed with respect to difficult real-world domains.
Files in this item
Google Scholar:Pérez, Eduardo
-
Rendell, Larry
This item appears in the following Collection(s)
Related items
Showing items related by title, author, creator and subject.
-
Constructive induction and genetic algorithms for learning concepts with complex interaction
Shafti, Leila Shila; Pérez, Eduardo
2005 -
Adipose-derived mesenchymal stromal cells for the treatment of patients with severe SARS-CoV-2 pneumonia requiring mechanical ventilation. A proof of concept study
Sánchez-Guijo, Fermín; García Arranz, Mariano Andrés; López-Parra, Miriam; Monedero, Pablo; Mata-Martínez, Carmen; Santos, Arnoldo; Sagredo, Víctor; Alvarez-Avello, José Manuel; Guerrero, José Eugenio; Pérez-Calvo, César; Sánchez-Hernández, Miguel Vicente; Del-Pozo, José Luis; Andreu, Enrique J.; Fernández-Santos, María Eugenia; Soria-Juan, Barbara; Hernández-Blasco, Luis M.; Andreu, Etelvina; Sempere, José M.; Zapata, Agustín G.; Moraleda, José M.; Soria, Bernat; Fernández-Avilés, Francisco; García Olmo, Damián; Prósper, Felipe
2020-07-10