Algorithm selection on data streams

J.R. Rijn, van, G. Holmes, B. Pfahringer, J. Vanschoren

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

26 Citaten (Scopus)


We explore the possibilities of meta-learning on data streams, in particular algorithm selection. In a first experiment we calculate the characteristics of a small sample of a data stream, and try to predict which classifier performs best on the entire stream. This yields promising results and interesting patterns. In a second experiment, we build a meta-classifier that predicts, based on measurable data characteristics in a window of the data stream, the best classifier for the next window. The results show that this meta-algorithm is very competitive with state of the art ensembles, such as OzaBag, OzaBoost and Leveraged Bagging. The results of all experiments are made publicly available in an online experiment database, for the purpose of verifiability, reproducibility and generalizability.
Originele taal-2Engels
TitelDiscovery Science (17th International Conference, DS 2014, Bled, Slovenia, October 8-10, 2014. Proceedings)
RedacteurenS. Dzeroski, P. Panov, D. Kocev, L. Todorovski
Plaats van productieHeidelberg
ISBN van geprinte versie978-3-319-11811-6
StatusGepubliceerd - 2014
Evenementconference; 17th International Conference on Discovery Science; 2014-10-08; 2014-10-10 -
Duur: 8 okt 201410 okt 2014

Publicatie series

NaamLecture Notes in Computer Science
ISSN van geprinte versie0302-9743


Congresconference; 17th International Conference on Discovery Science; 2014-10-08; 2014-10-10
Ander17th International Conference on Discovery Science

Vingerafdruk Duik in de onderzoeksthema's van 'Algorithm selection on data streams'. Samen vormen ze een unieke vingerafdruk.

Citeer dit