Geodesic
k-Best Max-margin Approaches for Sequence Labeling
Computer Science and Information Systems, Tome 12 (2015) no. 2.

Voir la notice de l'article provenant de la source Computer Science and Information Systems website

Structured learning algorithms usually require inference during the training procedure. Due to their exponential size of output space, the parameter update is performed only on a relatively small collection built from the “best” structures. The k-best MIRA is an example of an online algorithm which seeks optimal parameters by making updates on k structures with the highest score at a time. Following the idea of using k-best structures during the learning process, in this paper we introduce four new k-best extensions of max-margin structured algorithms. We discuss their properties and connection, and evaluate all algorithms on two sequence labeling problems, the shallow parsing and named entity recognition. The experiments show how the proposed algorithms are affected by the changes of k in terms of the F-measure and computational time, and that the proposed algorithms can improve results in comparison to the single best case. Moreover, the restriction to the single best case produces a comparison of the existing algorithms.
Keywords: structured learning, sequence labeling, k-best approach, max-margin training 
@article{CSIS_2015_12_2_a8,
     author = {Dejan Man\v{c}ev and Branimir Todorovi\'c},
     title = {k-Best {Max-margin} {Approaches} for {Sequence} {Labeling}},
     journal = {Computer Science and Information Systems},
     publisher = {mathdoc},
     volume = {12},
     number = {2},
     year = {2015},
     url = {http://geodesic.mathdoc.fr/item/CSIS_2015_12_2_a8/}
}
TY  - JOUR
AU  - Dejan Mančev
AU  - Branimir Todorović
TI  - k-Best Max-margin Approaches for Sequence Labeling
JO  - Computer Science and Information Systems
PY  - 2015
VL  - 12
IS  - 2
PB  - mathdoc
UR  - http://geodesic.mathdoc.fr/item/CSIS_2015_12_2_a8/
ID  - CSIS_2015_12_2_a8
ER  - 
%0 Journal Article
%A Dejan Mančev
%A Branimir Todorović
%T k-Best Max-margin Approaches for Sequence Labeling
%J Computer Science and Information Systems
%D 2015
%V 12
%N 2
%I mathdoc
%U http://geodesic.mathdoc.fr/item/CSIS_2015_12_2_a8/
%F CSIS_2015_12_2_a8
Dejan Mančev; Branimir Todorović. k-Best Max-margin Approaches for Sequence Labeling. Computer Science and Information Systems, Tome 12 (2015) no. 2. http://geodesic.mathdoc.fr/item/CSIS_2015_12_2_a8/