Abstract：To determine the chain order of the multi-label classifier chain algorithm and mine the high-level label corelation, a multi-label classifier chain method was proposed based on gradient boosting. The overall framework of GBCC was given, and the gradient boosting of a single label was conducted by pre-pruning strategy. In the process, the label confidence and error evaluation scores were used to determine the optimal chain order, and the label transfer and feature transfer were performed among labels to mine the high-level label relevance. The proposed method was compared with 4 classifier chain algorithms of CC, ECC, OCC and EOCC and 4 multi-label classification algorithms of BR, HOMER, MLKNN and CLR on 12 multi-label data sets such as bibtex and Corel5k. The results show that the new method can effectively improve the prediction performance under various evaluation indicators of micro-F1, macro-F1, Hamming loss and One-error with maintaining the flexibility of classifier chain algorithm.
ZHENG H Z, ZUO W L. Multi-labeled social networks users personality prediction based on information gain and semantic features\[J\]. Journal of Jilin University（Science Edition）, 2016, 54(3): 561-568.（in Chinese）
DENG K J, HUA K, DENG C M,et al. Research on author name disambiguation method based on machine learning\[J\]. Journal of Sichuan University (Natural Science Edition), 2019, 56(2): 241-245. （in Chinese）
［3］
WANG J, YANG Y, MAO J H, et al. CNN-RNN: a unified framework for multi-label image classification\[C\]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Computer Society, 2016: 2285-2294.
［4］
CHENG X, ZHAO S G, XIAO X, et al. iATC-mISF: a multi-label classifier for predicting the classes of anato-mical therapeutic chemicals\[J\]. Bioinformatics, 2017, 33(3): 341-346.
［5］
ZHANG M L, ZHOU Z H. A review on multi-label learning algorithms\[J\]. IEEE Transactions on Know-ledge and Data Engineering, 2014, 26(8): 1819-1837.
［6］
READ J, PFAHRINGER B, HOLMES G, et al. Classifier chains for multi-label classification\[C\]∥Proceedings of the Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Heidelberg: Springer, 2009: 254-269.
［7］
PROKHORENKOVA L, GUSEV G, VOROBEV A, et al. CatBoost: unbiased boosting with categorical features\[C\]∥Proceedings of the 2018 Conference on Advances in Neural Information Processing Systems. Montreal: NeurIPS, 2018: 6638-6648.
WANG J, CHEN Z L, LI H, et al. Hierarchical multi-label classification using incremental hypernetwork \[J\]. Journal of Chongqing University of Posts and Telecommunications (Natural Science Edition), 2019, 31(4): 538-549.（in Chinese）
［9］
NARASSIGUIN A, ELGHAZEL H, AUSSEM A. Dynamic ensemble selection with probabilistic classifier chains\[C\]∥Proceedings of the Joint European Confe-rence on Machine Learning and Knowledge Discovery in Databases. Macedonia: Springer, 2017: 169-186.
［10］
JIMNEZ R B, MORALES E F, ESCALANTE H J. Bayesian chain classifier with feature selection for multi-label classification\[C\]∥Proceedings of the Mexican International Conference on Artificial Intelligence. Mexico: Springer, 2018: 232-243.
［11］
SUN L, KUDO M. Optimization of classifier chains via conditional likelihood maximization\[J\]. Pattern Recognition, 2018, 74: 503-517.
［12］
DEMAR J. Statistical comparisons of classifiers over multiple data sets\[J\]. Journal of Machine Learning Research, 2006, 7(1): 1-30.
2021, Vol. 42 
