www-ai.cs.tu-dortmund.de/de/LEHRE/SEMINARE/SS09/AKTARBEITENDESDM/LITERATUR/SequenceLabelling.pdf
LNAI 5211 - Sequence Labelling SVMs Trained in One Pass
600) 5500 (%43,000) 26 128 0.1 5 10 10 1 Chunking 8,931 (%212,000) 2,012 (%47,000) 21 %76,000 0.1 1 2 5 1 WSJ 42,466 (%1,000,000) 2,155 (%53,000) 44 %130,000 0.1 1 2 5 1
Table 1 summarizes the main chara [...] the se- quence of tokens x = (x1. . . xT ) or the sequence of labels y = (y1. . . yT ). Subse- quences are denoted using superscripts, as in y{t"k..t"1} = (yt"k. . . yt"1). We call X the set of possible [...] y{t"k..t"1}, yt
#$ t = 1...T ,
where w $ RD is a parameter vector and function !g : X % Yk % Y & RD de- termines the feature space.
2.1 Exact Inference
Exact inference maximizes the sum %T
t=1 st(w,x,y) …
