Czech
English
You are here
Bibliography
Conference Paper (international conference)
Minimum Description Length Principle for Compositional Model Learning
,
: Integrated Uncertainty in Knowledge Modelling and Decision Making, p. 254-266 , Eds: Huynh Van-Nam, Inuiguchi Masahiro, Denoeux Thierry
: 4th Intemational Symposium, IUKM 2015, (Nha Trang, VN, 20151015)
: GA15-00215S, GA ČR
: Machine learning, Multidimensional models, Probability distributions, Composition, Information theory, Lossless encoding
: 10.1007/978-3-319-25135-6_25
: http://library.utia.cas.cz/separaty/2019/MTR/jirousek-0507131.pdf
(eng): Not having another source of information than source data, the process of data-based model construction can be viewed as the transformation of information represented by data into that represented by the model. The paper explains how this idea supports the Minimum Description Length Principle and how it can be employed to avoid the overfitting of the constructed model.
: IN
: 10201