Author | dc.contributor.author | Silva, Jorge F. | |
Admission date | dc.date.accessioned | 2012-08-14T21:00:28Z | |
Available date | dc.date.available | 2012-08-14T21:00:28Z | |
Publication date | dc.date.issued | 2012-03 | |
Cita de ítem | dc.identifier.citation | IEEE TRANSACTIONS ON INFORMATION THEORY Volume: 58 Issue: 3 Pages: 1940-1952 Published: MAR 2012 | es_CL |
Identifier | dc.identifier.other | DOI: 10.1109/TIT.2011.2177771 | |
Identifier | dc.identifier.uri | https://repositorio.uchile.cl/handle/2250/125678 | |
General note | dc.description | Artículo de publicación ISI | es_CL |
Abstract | dc.description.abstract | A new histogram-based mutual information estimator using data-driven tree-structured partitions (TSP) is presented in this paper. The derived TSP is a solution to a complexity regularized empirical information maximization, with the objective of finding a good tradeoff between the known estimation and approximation errors. A distribution-free concentration inequality for this tree-structured learning problem as well as finite sample performance bounds for the proposed histogram-based solution is derived. It is shown that this solution is density-free strongly consistent and that it provides, with an arbitrary high probability, an optimal balance between the mentioned estimation and approximation errors. Finally, for the emblematic scenario of independence, I(X;Y), it is shown that the TSP estimate converges to zero with O(e(-n1/3+log log n)). | es_CL |
Patrocinador | dc.description.sponsorship | National Commission for Scientific and Technological Research (CONICYT), Chile, under FONDECYT
1110145 | es_CL |
Lenguage | dc.language.iso | en | es_CL |
Keywords | dc.subject | Complexity regularization | es_CL |
Título | dc.title | Complexity-Regularized Tree-Structured Partition for Mutual Information Estimation | es_CL |
Document type | dc.type | Artículo de revista | |