Korbicz, Józef - red. ; Uciński, Dariusz - red.
We present a primal sub-gradient method for structured SVM optimization defined with the averaged sum of hinge losses inside each example. Compared with the mini-batch version of the Pegasos algorithm for the structured case, which deals with a single structure from each of multiple examples, our algorithm considers multiple structures from a single example in one update. This approach should increase the amount of information learned from the example. ; We show that the proposed version with the averaged sum loss has at least the same guarantees in terms of the prediction loss as the stochastic version. Experiments are conducted on two sequence labeling problems, shallow parsing and part-of-speech tagging, and also include a comparison with other popular sequential structured learning algorithms.
Zielona Góra: Uniwersytet Zielonogórski
AMCS, volume 24, number 4 (2014) ; kliknij tutaj, żeby przejść
Biblioteka Uniwersytetu Zielonogórskiego
2024-04-29
2024-04-29
21
https://zbc.uz.zgora.pl/publication/88828
Nazwa wydania | Data |
---|---|
A primal sub-gradient method for structured classification with the averaged sum loss | 2024-04-29 |
Yuan, Liming Liu, Jiafeng Tang, Xianglong Abaev, Pavel - ed. Razumchik, Rostislav - ed. Kołodziej, Joanna - ed.
Yao, Baozhen Hu, Ping Zhang, Mingheng Jin, Maoqing Makowski, Ryszard - ed. Zarzycki, Jan - ed.
Chmielnicki, Wiesław Stąpor, Katarzyna Korbicz, Józef - red. Uciński, Dariusz - red.
Fujarewicz, Krzysztof Wiench, Małgorzata Kimmel, Marek - red. Lachowicz, Mirosław - red. Świerniak, Andrzej - red.
Rybka, Jan Janicki, Artur Korbicz, Józef - red. Uciński, Dariusz - red.
Bilski, Adrian Wojciechowski, Jacek Korbicz, Józef - red. Uciński, Dariusz - red.
Juszczyk, Michał Kuczyński, Tadeusz - red.