Label correlation for partial label learning

doi:10.23919/JSEE.2022.000102

Journal of Systems Engineering and Electronics ›› 2022, Vol. 33 ›› Issue (5): 1043-1051.doi: 10.23919/JSEE.2022.000102

• ELECTRONICS TECHNOLOGY • Previous Articles Next Articles

Label correlation for partial label learning

Lingchi GE(), Min FANG*(), Haikun LI(), Bo CHEN()

¹ School of Computer Science and Technology, Xidian University, Xi’an 710071, China

Received:2021-05-10 Online:2022-10-27 Published:2022-10-27
Contact: Min FANG E-mail:15109285306@163.com;fanglabtg@163.com;haikun1990@163.com;bchen_0314@stu.xidian.edu.cn
About author:|GE Lingchi was born in 1997. He received his B.S. degree from Chang’an University, Xi’an, China, in 2018. He is currently pursuing his M.S. degree in computer science at Xidian University, Xi’an, China. His research interests are pattern recognition, partial label learning, and multi-label learning. E-mail: 15109285306@163.com||FANG Min was born in 1965. She received her B.S. degree in computer control, M.S. degree in computer software engineering, and Ph.D. degree in computer application from Xidian University, Xi’an, China, in 1986, 1991, and 2004, respectively, where she is currently a professor. Her research interests include intelligent information process, multi-agent system, and network technology. E-mail: fanglabtg@163.com||LI Haikun was born in 1990. He received his M.S. degree from Yunnan University, Kunming, China, in 2017. He is currently a Ph.D. candidate in technology of computer application at Xidian University, Xi’an, China. His research interests include pattern recognition, machine learning, and partial-label learning. E-mail: haikun1990@163.com||CHEN Bo was born in 1993. He received his B.S. degree in 2014 from Xi’an University of Post & Telecomunications, Xi’an, China. He is currently a Ph.D. candidate in computer science at Xidian University, Xi’an, China. His research interests include pattern recognition, machine learning, and time series. E-mail: bchen_0314@stu.xidian.edu.cn
Supported by:
This work was supported by the National Natural Science Foundation of China (62176197;61806155) and the National Natural Science Foundation of Shaanxi Province (2020GY-062).

Abstract

Abstract:

Partial label learning aims to learn a multi-class classifier, where each training example corresponds to a set of candidate labels among which only one is correct. Most studies in the label space have only focused on the difference between candidate labels and non-candidate labels. So far, however, there has been little discussion about the label correlation in the partial label learning. This paper begins with a research on the label correlation, followed by the establishment of a unified framework that integrates the label correlation, the adaptive graph, and the semantic difference maximization criterion. This work generates fresh insight into the acquisition of the learning information from the label space. Specifically, the label correlation is calculated from the candidate label set and is utilized to obtain the similarity of each pair of instances in the label space. After that, the labeling confidence for each instance is updated by the smoothness assumption that two instances should be similar outputs in the label space if they are close in the feature space. At last, an effective optimization program is utilized to solve the unified framework. Extensive experiments on artificial and real-world data sets indicate the superiority of our proposed method to state-of-art partial label learning methods.

Key words: pattern recognition, partial label learning, label correlation, disambiguation

Lingchi GE, Min FANG, Haikun LI, Bo CHEN. Label correlation for partial label learning[J]. Journal of Systems Engineering and Electronics, 2022, 33(5): 1043-1051.

Figures/Tables 6

Table 1

Table 2

Table 3

Table 4

Fig 1

Fig 2

References 30

1	LIU L P, DIETTERICH T G A conditional multinomial mixture model for superset label learning. Advances in Neural Information Processing Systems, 2012, 1, 548- 556.
2	ZHOU D, ZHANG Z, ZHANG M L, et al Weakly supervised POS tagging without disambiguation. ACM Transactions on Asian and Low-Resource Language Information Processing, 2018, 17 (4): 1- 19.
3	ZENG Z N, XIAO S J, JIA K, et al Learning by associating ambiguously labeled images. Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2013, 708- 715.
4	YAO Y, DENG J H, CHEN X H, et al Deep discriminative CNN with temporal ensembling for ambiguously-labeled image classification. Proc. of the AAAI Conference on Artificial Intelligence, 2020, 34 (7): 12669- 12676. doi: 10.1609/aaai.v34i07.6959
5	HULLERMEIER E, BERINGER J Learning from ambiguously labeled examples. Intelligent Data Analysis, 2006, 10 (5): 419- 439. doi: 10.3233/IDA-2006-10503
6	COUR T, SAPP B, TASKAR B Learning from partial labels. Journal of Machine Learning Research, 2011, 12, 1501- 1536.
7	NGUYEN N, CARUANA R Classification with partial labels. Proc. of the 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2008, 551- 559.
8	ZHANG M L, YU F Solving the partial label learning problem: an instance-based approach. Proc. of the International Joint Conference on Artificial Intelligence, 2015, 4048- 4054.
9	ZHANG M L, ZHOU B B, LIU X Y Partial label learning via feature-aware disambiguation. Proc. of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016, 1335- 1344.
10	GONG C, LIU T L, TANG Y Y, et al A regularization approach for instance-based superset label learning. IEEE Trans. on Cybernetics, 2017, 48 (3): 967- 978.
11	WANG D B, LI L, ZHANG M L Adaptive graph guided disambiguation for partial label learning. Proc. of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2019, 83- 91.
12	LYU G Y, FENG S H, WANG T, et al GM-PLL: graph matching based partial label learning. IEEE Trans. on Knowledge and Data Engineering, 2021, 33 (2): 521- 535. doi: 10.1109/TKDE.2019.2933837
13	ZHU Y, KWOK J T, ZHOU Z H Multi-label learning with global and local label correlation. IEEE Trans. on Knowledge and Data Engineering, 2017, 30 (6): 1081- 1094.
14	FENG L, AN B Partial label learning by semantic difference maximization. Proc. of the International Joint Conference on Artificial Intelligence, 2019, 2294- 2300.
15	JIN R, GHAHRAMANI Z Learning with multiple labels. Proc. of the Annual Conference on Neural Information Processing Systems, 2002, 2, 897- 904.
16	YU F, ZHANG M L Maximum margin partial label learning. Proc. of the Asian Conference on Machine Learning, 2016, 96- 111.
17	CHEN Y C, PATEL V M, CHELLAPPA R, et al Ambiguously labeled learning using dictionaries. IEEE Trans. on Information Forensics and Security, 2014, 9 (12): 2076- 2088. doi: 10.1109/TIFS.2014.2359642
18	TANG C Z, ZHANG M L Confidence-rated discriminative partial label learning. Proc. of the AAAI Conference on Artificial Intelligence, 2017, 31 (1): 2611- 1618.
19	LYU G Y, FENG S H, LI Y D, et al Hera: partial label learning by combining heterogeneous loss with sparse and low-rank regularization. ACM Trans. on Intelligent Systems and Technology, 2020, 11 (3): 1- 19.
20	ZHANG Y B, YANG G, ZHAO S Y, et al Partial label learning via generative adversarial nets. Proc. of the European Conference on Artificial Intelligence, 2020, 1674- 1681.
21	ZHANG M L, YU F, TANG C Z Disambiguation-free partial label learning. IEEE Trans. on Knowledge and Data Engineering, 2017, 29 (10): 2155- 2167. doi: 10.1109/TKDE.2017.2721942
22	WU X, ZHANG M L Towards enabling binary decomposition for partial label learning. Proc. of the International Joint Conference on Artificial Intelligence, 2018, 2868- 2874.
23	YUILLE A L, RANGARAJAN A The concave-convex procedure (CCCP). Advances in Neural Information Processing Systems, 2002, 2, 1033- 1040.
24	SRIPERUMBUDUR B K, LANCKRIET G R G On the convergence of the concave-convex procedure. Proc. of the Annual Conference on Neural Information Processing Systems, 2009, 9, 1759- 1767.
25	XU S, YANG M, ZHOU Y, et al Partial label metric learning by collapsing classes. International Journal of Machine Learning and Cybernetics, 2020, 11, 2453- 2460. doi: 10.1007/s13042-020-01129-z
26	LYU J Q, XU M, FENG L, et al Progressive identification of true labels for partial-label learning. Proc. of the International Conference on Machine Learning, 2020, 6500- 6510.
27	YAN Y, GUO Y H Partial label learning with batch label correction. Proc. of the AAAI Conference on Artificial Intelligence, 2020, 34 (4): 6575- 6582. doi: 10.1609/aaai.v34i04.6132
28	PANIS G, LANITIS A, TSAPATSOULIS N, et al Overview of research on facial ageing using the FG-NET ageing database. IET Biometrics, 2016, 5 (2): 37- 46. doi: 10.1049/iet-bmt.2014.0053
29	BRIGGS F, FERN X Z, RAICH R Rank-loss support instance machines for MIML instance annotation. Proc. of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2012, 534- 542.
30	GUILLAUMIN M, VERBEEK J, SCHMID C Multiple instance metric learning from automatically labeled bags of faces. European Conference on Computer Vision, 2010, 634- 647.

Data set	Instance	Feature	PCA	Class
Enron	196	623	?	7
Medical	752	1287	?	30
Scene	2230	294	?	6
Bibtex	2744	1894	800	89
Mediamill	2854	120	?	10
Tmc2007	3130	15418	1000	21

Data set	Example	Feature	Class	Avg. CLs	Task domain
FG-NET	1002	262	78	7.48	Facial age estimation [28]
Lost	1122	108	16	2.23	Automatic face naming [6]
MSRCv2	1758	48	23	3.16	Object classification [1]
BirdSong	4998	38	13	2.18	Bird song classification [29]
Soccer Player	17472	279	171	2.09	Automatic face naming [3]
Yahoo! News	22991	163	219	1.91	Automatic face naming [30]

Data set	PL-LCSA	SDIM	PL-AGGD	PL-ECOC	IPAL	PL-KNN
Lost	0.789±0.030	0.797±0.030	0.744±0.020 $ \bullet $	0.706±0.043 $ \bullet $	0.645±0.034 $ \bullet $	0.459±0.039 $ \bullet $
MSRCv2	0.562±0.021	0.500±0.023 $ \bullet $	0.509±0.028 $ \bullet $	0.427±0.024 $ \bullet $	0.531±0.037 $ \bullet $	0.418±0.046 $ \bullet $
BirdSong	0.756±0.008	0.734±0.012 $ \bullet $	0.734±0.009 $ \bullet $	0.751±0.013	0.712±0.015 $ \bullet $	0.603±0.013 $ \bullet $
Soccer Player	0.596±0.013	0.577±0.016 $ \bullet $	0.539±0.016 $ \bullet $	0.169±0.005 $ \bullet $	0.544±0.014 $ \bullet $	0.494±0.012 $ \bullet $
Yahoo! News	0.670±0.008	0.663±0.013	0.647±0.009 $ \bullet $	0.561±0.011 $ \bullet $	0.607±0.012 $ \bullet $	0.471±0.005 $ \bullet $
FG-NET	0.079±0.020	0.076±0.037	0.076±0.027	0.005±0.007 $ \bullet $	0.061±0.018 $ \bullet $	0.066±0.018

Data set	$\delta = 0.5$ , $\gamma = 0.5$	$\delta = 0.5$ , $\gamma = 0$	$\delta = 0$ , $\gamma = 0$
Lost	0.789±0.030	0.767±0.032	0.746±0.025
MSRCv2	0.562±0.021	0.544±0.030	0.507±0.028
BirdSong	0.756±0.008	0.751±0.010	0.733±0.011
Soccer Player	0.596±0.013	0.589±0.013	0.538±0.015
Yahoo! News	0.670±0.008	0.666±0.009	0.648±0.009
FG-NET	0.079±0.020	0.071±0.023	0.077±0.024

[1]	Jujie Zhang, Min Fang, and Huimin Chai. Multi-label local discriminative embedding [J]. Systems Engineering and Electronics, 2017, 28(5): 1009-1018.
[2]	Jun He and Qiang Fu. Posterior probability calculation procedure for recognition rate comparison [J]. Systems Engineering and Electronics, 2016, 27(3): 700-711.
[3]	Xue Zhenxia, Liu Sanyang & Liu Wanli. Progressive transductive learning pattern classification via single sphere [J]. Journal of Systems Engineering and Electronics, 2009, 20(3): 643-650.
[4]	Chen Aijun, Li Jinzong & Zhu Bing. Circular object recognition based on shape parameters [J]. Journal of Systems Engineering and Electronics, 2007, 18(2): 199-204.

Label correlation for partial label learning

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

Share this article

Figures/Tables 6

References 30

Related Articles 4

Recommended Articles

Metrics

Comments