{"type":"ScholarlyArticle","title":"Does Tail Label Help for Large-Scale Multi-Label Learning","authors":[{"name":"Tong Wei","slug":"tong-wei","given":"Tong","family":"Wei","affiliation":"National Key Laboratory for Novel Software Technology , Nanjing University Collaborative Innovation Center of Novel Software Technology and Industrialization Nanjing 210023 , China"},{"name":"Yu-Feng Li","slug":"yu-feng-li","given":"Yu-Feng","family":"Li","affiliation":"National Key Laboratory for Novel Software Technology , Nanjing University Collaborative Innovation Center of Novel Software Technology and Industrialization Nanjing 210023 , China"}],"abstract":"Large-scale multi-label learning annotates relevant labels for unseen data from a huge number of candidate labels. It is well known that in large-scale multi-label learning, labels exhibit a long tail distribution in which a significant fraction of labels are tail labels. Nonetheless, how tail labels make impact on the performance metrics in large-scale multi-label learning was not explicitly quantified. In this paper, we disclose that whatever labels are randomly missing or misclassified, tail labels impact much less than common labels in terms of commonly used performance metrics (Top-$k$ precision and nDCG@$k$). With the observation above, we develop a low-complexity large-scale multi-label learning algorithm with the goal of facilitating fast prediction and compact models by trimming tail labels adaptively. Experiments clearly verify that both the prediction time and the model size are significantly reduced without sacrificing much predictive performance for state-of-the-art approaches.","doi":"10.24963/ijcai.2018/395","url":"https://ijcai.org/proceedings/2018/395","pdf":"https://ijcai.org/proceedings/2018/0395.pdf","markdown":"https://ijcai.org/proceedings/2018/395.md","bibtex":"https://ijcai.org/proceedings/2018/bibtex/395","volume":{"slug":"2018","title":"Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2018","url":"https://ijcai.org/proceedings/2018/","year":2018,"location":"Stockholm","dates":"13-19 July 2018","editors":"Jérôme Lang","isbn":"978-0-9992411-2-7","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2018"},"section":"Machine Learning","track":"Main track","number":395,"pages":{"first":2847,"last":2853},"keywords":["Machine Learning Applications: Big data ; Scalability","Machine Learning: Multi-instance;Multi-label;Multi-view learning"],"videos":[],"references":[{"position":1,"raw":"[Babbar and Schölkopf, 2017] R. Babbar and B. Schölkopf. Dismec: Distributed sparse machines for extreme multi- label classification. In Proceedings of the 10th ACM In- ternational Conference on Web Search and Data Mining, pages 721-729, Cambridge, UK, 2017.","title":"Dismec: Distributed sparse machines for extreme multilabel classification","authors":"R Babbar, B Schölkopf","year":"2017","venue":"Proceedings of the 10th ACM In-ternational Conference on Web Search and Data Mining","doi":null,"url":null,"internal_url":null},{"position":2,"raw":"[Babbar and Schölkopf, 2018] R. Babbar and B. Schölkopf. Adversarial extreme multi-label classification. arXiv preprint arXiv:1803.01570, 2018.","title":"Adversarial extreme multi-label classification","authors":"R Babbar, B Schölkopf","year":"2018","venue":null,"doi":null,"url":null,"internal_url":null},{"position":3,"raw":"[Bartlett, 1998] P. L. Bartlett. The sample complexity of pat- tern classification with neural networks: the size of the weights is more important than the size of the network. IEEE Transactions on Information Theory, 44(2):525- 536, 1998.","title":"The sample complexity of pattern classification with neural networks: the size of the weights is more important than the size of the network","authors":"P L Bartlett","year":"1998","venue":"IEEE Transactions on Information Theory","doi":null,"url":null,"internal_url":null},{"position":4,"raw":"[Bhatia et al., 2015] K. Bhatia, H. Jain, P. Kar, M. Varma, and P. Jain. Sparse local embeddings for extreme multi- label classification. In Advances in Neural Information Processing Systems, pages 730-738, Montreal, Canada, 2015.","title":"Sparse local embeddings for extreme multilabel classification","authors":"K Bhatia, H Jain, P Kar, M Varma, P Jain","year":"2015","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":5,"raw":"[Bi and Kwok, 2013] W. Bi and J. T. Kwok. Efficient multi- label classification with many labels. In Proceedings of the 30th International Conference on Machine Learning, pages 405-413, Atlanta, GA, 2013.","title":"Efficient multilabel classification with many labels","authors":"W Bi, J T Kwok","year":"2013","venue":"Proceedings of the 30th International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":6,"raw":"[Brent, 2013] R. P. Brent. Algorithms for minimization with- out derivatives. Courier Corporation, 2013.","title":"Algorithms for minimization without derivatives","authors":"R P Brent","year":"2013","venue":null,"doi":null,"url":null,"internal_url":null},{"position":7,"raw":"[Chen and Lin, 2012] Y.-N. Chen and H.-T. Lin. Feature- aware label space dimension reduction for multi-label classification. In Advances in Neural Information Process- ing Systems, pages 1529-1537, Lake Tahoe, NV, 2012.","title":"Featureaware label space dimension reduction for multi-label classification","authors":"Y.-N Chen, H.-T Lin","year":"2012","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":8,"raw":"A. E. Choromanska and J. Langford. Logarithmic time online multiclass predic- tion. In Advances in Neural Information Processing Sys- tems, pages 55-63, Montreal, Canada, 2015.","title":"Logarithmic time online multiclass prediction","authors":"A E Choromanska, J Langford","year":"2015","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":9,"raw":"[Cisse et al., 2013] M. M. Cisse, N. Usunier, T. Artieres, and P. Gallinari. Robust bloom filters for large multilabel classification tasks. In Advances in Neural Information Processing Systems, pages 1851-1859, Lake Tahoe, NV, 2013.","title":"Robust bloom filters for large multilabel classification tasks","authors":"M M Cisse, N Usunier, T Artieres, P Gallinari","year":"2013","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":10,"raw":"[Daume III et al., 2016] H. Daume III, N. Karampatziakis, J. Langford, and P. Mineiro. Logarithmic time one- against-some. arXiv preprint arXiv:1606.04988, 2016.","title":"Logarithmic time oneagainst-some","authors":"H Daume, N Karampatziakis, J Langford, P Mineiro","year":"2016","venue":null,"doi":null,"url":null,"internal_url":null},{"position":11,"raw":"[Forsythe et al., 1977] G. E. Forsythe, C. B. Moler, and M. A. Malcolm. Computer methods for mathematical computations. 1977.","title":"Computer methods for mathematical computations","authors":"G E Forsythe, C B Moler, M A Malcolm","year":"1977","venue":null,"doi":null,"url":null,"internal_url":null},{"position":12,"raw":"[Hsu et al., 2009] D. J. Hsu, S. M. Kakade, J. Langford, and T. Zhang. Multi-label prediction via compressed sensing. In Advances in Neural Information Processing Systems, pages 772-780, Vancouver, Canada, 2009.","title":"Multi-label prediction via compressed sensing","authors":"D J Hsu, S M Kakade, J Langford, T Zhang","year":"2009","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":13,"raw":"[Jain et al., 2016] H. Jain, Y. Prabhu, and M. Varma. Ex- treme multi-label loss functions for recommendation, tag- ging, ranking & other missing label applications. In Pro- ceedings of the 22nd ACM SIGKDD International Con- ference on Knowledge Discovery and Data Mining, pages 935-944, San Francisco, CA, 2016.","title":"Extreme multi-label loss functions for recommendation, tagging, ranking & other missing label applications","authors":"H Jain, Y Prabhu, M Varma","year":"2016","venue":"Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining","doi":null,"url":null,"internal_url":null},{"position":14,"raw":"[Li et al., 2017] J.-J. Li, K. Lu, Z. Huang, and H.-T. Shen. Two birds one stone: On both cold-start and long-tail rec- ommendation. In Proceedings of the 2017 ACM on Multi- media Conference, pages 898-906. ACM, 2017.","title":"Two birds one stone: On both cold-start and long-tail recommendation","authors":"J.-J Li, K Lu, Z Huang, H.-T Shen","year":"2017","venue":"Proceedings of the 2017 ACM on Multimedia Conference","doi":null,"url":null,"internal_url":null},{"position":15,"raw":"[Lim et al., 2015] D. Lim, J. McAuley, and G. Lanckriet. Top-n recommendation with missing implicit feedback. In Proceedings of the 9th ACM Conference on Recommender Systems, pages 309-312. ACM, 2015.","title":"Top-n recommendation with missing implicit feedback","authors":"D Lim, J Mcauley, G Lanckriet","year":"2015","venue":"Proceedings of the 9th ACM Conference on Recommender Systems","doi":null,"url":null,"internal_url":null},{"position":16,"raw":"[Lin et al., 2014] Z.-J. Lin, G.-G. Ding, M.-Q. Hu, and J.-M. Wang. Multi-label classification via feature-aware implicit label space encoding. In Proceedings of the 31st Interna- tional Conference on Machine Learning, pages 325-333, Beijing, China, 2014.","title":"Multi-label classification via feature-aware implicit label space encoding","authors":"Z.-J Lin, G.-G Ding, M.-Q Hu, J.-M Wang","year":"2014","venue":"Proceedings of the 31st International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":17,"raw":"[Prabhu and Varma, 2014] Y. Prabhu and M. Varma. Fastxml: A fast, accurate and stable tree-classifier for extreme multi-label learning. In Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 263-272, New York City, NY, 2014.","title":"Fastxml: A fast, accurate and stable tree-classifier for extreme multi-label learning","authors":"Y Prabhu, M Varma","year":"2014","venue":"Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining","doi":null,"url":null,"internal_url":null},{"position":18,"raw":"[Schapire, 1990] R. E. Schapire. The strength of weak learn- ability. Machine Learning, 5(2):197-227, 1990.","title":"The strength of weak learnability","authors":"R E Schapire","year":"1990","venue":"Machine Learning","doi":null,"url":null,"internal_url":null},{"position":19,"raw":"[Tai and Lin, 2012] F. Tai and H.-T. Lin. Multilabel classi- fication with principal label space transformation. Neural Computation, 24(9):2508-2542, 2012.","title":"Multilabel classification with principal label space transformation","authors":"F Tai, H.-T Lin","year":"2012","venue":"Neural Computation","doi":null,"url":null,"internal_url":null},{"position":20,"raw":"[Wang et al., 2017] Y.-X. Wang, D. Ramanan, and M. Hebert. Learning to model the tail. In Advances in Neural Information Processing Systems, pages 7032- 7042, 2017.","title":"Learning to model the tail","authors":"Y.-X Wang, D Ramanan, M Hebert","year":"2017","venue":"Advances in Neural Information Processing Systems","doi":null,"url":null,"internal_url":null},{"position":21,"raw":"[Xu et al., 2016] C. Xu, D.-C. Tao, and C. Xu. Robust ex- treme multi-label learning. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 1275-1284, San Fran- cisco, CA, 2016.","title":"Robust extreme multi-label learning","authors":"C Xu, D.-C Tao, C Xu","year":"2016","venue":"Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining","doi":null,"url":null,"internal_url":null},{"position":22,"raw":"[Yeh et al., 2017] C.-K. Yeh, W.-C. Wu, W.-J. Ko, and Y.- C. F. Wang. Learning deep latent space for multi-label classification. In Proceedings of the 31st AAAI Conference on Artificial Intelligence, pages 2838-2844, San Fran- cisco, CA, 2017.","title":"Learning deep latent space for multi-label classification","authors":"C.-K Yeh, W.-C Wu, W.-J Ko, Y.-C F Wang","year":"2017","venue":"Proceedings of the 31st AAAI Conference on Artificial Intelligence","doi":null,"url":null,"internal_url":null},{"position":23,"raw":"[Yen et al., 2016] I. E.-H. Yen, X.-R. Huang, P. Ravikumar, K. Zhong, and I. S. Dhillon. Pd-sparse: A primal and dual sparse approach to extreme multiclass and multilabel classification. In Proceedings of the 33rd International Conference on Machine Learning, pages 3069-3077, New York City, NY, 2016.","title":"Pd-sparse: A primal and dual sparse approach to extreme multiclass and multilabel classification","authors":"Yen","year":"2016","venue":"Proceedings of the 33rd International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":24,"raw":"[Yu et al., 2014] H.-F. Yu, P. Jain, P. Kar, and I. S. Dhillon. Large-scale multi-label learning with missing labels. In Proceedings of the 31st International Conference on Ma- chine Learning, pages 593-601, Beijing, China, 2014.","title":"Large-scale multi-label learning with missing labels","authors":"H.-F Yu, P Jain, P Kar, I S Dhillon","year":"2014","venue":"Proceedings of the 31st International Conference on Machine Learning","doi":null,"url":null,"internal_url":null},{"position":25,"raw":"[Zhang and Schneider, 2011] Y. Zhang and J. Schneider. Multi-label output codes using canonical correlation anal- ysis. In Proceedings of the 14th International Conference on Artificial Intelligence and Statistics, pages 873-882, Ft.","title":"Multi-label output codes using canonical correlation analysis","authors":"Y Zhang, J Schneider","year":null,"venue":"Proceedings of the 14th International Conference on Artificial Intelligence and Statistics","doi":null,"url":null,"internal_url":null},{"position":26,"raw":"[Zhang and Zhou, 2014] M.-L. Zhang and Z.-H. Zhou. A re- view on multi-label learning algorithms. IEEE Transac- tions on Knowledge and Data Engineering, 26(8):1819- 1837, 2014.","title":"A review on multi-label learning algorithms","authors":"M.-L Zhang, Z.-H Zhou","year":"2014","venue":"IEEE Transactions on Knowledge and Data Engineering","doi":null,"url":null,"internal_url":null}],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}