Metric learning with geometric mean for similarities measurement (original) (raw)

Abstract

Distance metric learning aims to find an appropriate method to measure similarities between samples. An excellent distance metric can greatly improve the performance of many machine learning algorithms. Most previous methods in this area have focused on finding metrics which utilize large-margin criterion to optimize compactness and separability simultaneously. One major shortcoming of these methods is their failure to balance all between-class scatters when the distributions of samples are extremely unbalanced. Large-margin criterion tends to maintain bigger scatters while abandoning those smaller ones to make the total scatters maximized. In this paper, we introduce a regularized metric learning framework, metric learning with geometric mean which obtains a distance metric using geometric mean. The novel method balances all between-class scatters and separates samples from different classes simultaneously. Various experiments on benchmark datasets show the good performance of the novel method.

Access this article

Log in via an institution

Subscribe and save

• Starting from 10 chapters or articles per month
• Access and download chapters and articles from more than 300k books and 2,500 journals
• Cancel anytime View plans

Buy Now

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

Notes

References

• Baghshah M S, Shouraki SB (2009) Semi-supervised metric learning using pairwise constraints [C]. In: IJCAI, vol 9, pp 1217–1222
• Cover TM, Thomas JA (2012) Elements of information theory [M]. Wiley, New York
  Google Scholar
• Davis JV, Dhillon IS (2008) Structured metric learning for high dimensional problems [C]. In: Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining. ACM, pp 195–203
• Davis JV, Kulis B, Jain P et al (2007) Information-theoretic metric learning [C]. In: Proceedings of the 24th international conference on machine learning. ACM, pp 209–216
• Duda RO, Hart PE, Stork DG (2012) Pattern classification [M]. Wiley, New York
  MATH Google Scholar
• Feng L, Wang H, Liu S et al (2015) Locality structured sparsity preserving embedding [J]. Int J Pattern Recognit Artif Intell 29(06):1551010
  Article Google Scholar
• Goldberger J, Hinton GE, Roweis ST et al (2004) Neighbourhood components analysis. In: Advances in neural information processing systems, pp 513–520
• Hoi SC, Liu W, Chang SF (2010) Semi-supervised distance metric learning for collaborative image retrieval and clustering [J]. ACM Trans Multimed Comput Commun Appl (TOMM) 6(3):18
  Google Scholar
• Huang K, Ying Y, Campbell C (2011) Generalized sparse metric learning with relative comparisons [J]. Knowl Inf Syst 28(1):25–45
  Article Google Scholar
• Lee JH, McDonnell KT, Zelenyuk A et al (2014) A structure-based distance metric for high-dimensional space exploration with multidimensional scaling [J]. Vis Comput Gr IEEE Trans 20(3):351–364
  Article Google Scholar
• Liu W, Tian X, Tao D et al (2010) Constrained metric learning via distance gap maximization [C]. In: AAAI
• Ma L, Yang X, Tao D (2014) Person re-identification over camera networks using multi-task distance metric learning [J]. Image Process IEEE Trans 23(8):3656–3670
  Article MathSciNet Google Scholar
• Niyogi X (2004) Locality preserving projections [C]. In: Neural information processing systems, vol 16. MIT, p 153
• Noorbehbahani F, Mousavi SR, Mirzaei A (2015) An incremental mixed data clustering method using a new distance measure [J]. Soft Comput 19(3):731–743
  Article Google Scholar
• Roweis ST, Saul LK (2000) Nonlinear dimensionality reduction by locally linear embedding [J]. Science 290(5500):2323–2326
  Article Google Scholar
• Schölkopf B, Smola AJ (2002) Learning with kernels: support vector machines, regularization, optimization, and beyond [M]. MIT press, Cambridge
  Google Scholar
• Shao C, Song X, Yang X et al (2015) Extended minimum-squared error algorithm for robust face recognition via auxiliary mirror samples [J]. Soft Comput. doi:10.1007/s00500-015-1692-7
• Shental N, Hertz T, Weinshall D (2002) Adjustment learning and relevant component analysis [M]. In: Computer vision-ECCV 2002. Springer. Berlin, Heidelberg, pp 776–790
• Tao D, Li X, Wu X et al (2009) Geometric mean for subspace selection [J]. Pattern Anal Mach Intell IEEE Trans 31(2):260–274
  Article Google Scholar
• Wang F (2011) Semisupervised metric learning by maximizing constraint margin [J]. Syst Man Cybern Part B: Cybern IEEE Trans 41(4):931–939
  Article Google Scholar
• Wang S, Jin R (2009) An information geometry approach for distance metric learning [C]. In: International conference on artificial intelligence and statistics, pp 591–598
• Wei J, Zeng Q, Wang X et al (2014) Integrating local and global topological structures for semi-supervised dimensionality reduction [J]. Soft Comput 18(6):1189–1198
  Article Google Scholar
• Weinberger KQ, Blitzer J, Saul LK (2005) Distance metric learning for large margin nearest neighbor classification [C]. In: Advances in neural information processing systems, pp 1473–1480
• Wold S, Esbensen K, Geladi P (1987) Principal component analysis [J]. Chemometr Intell Lab Syst 2(1):37–52
  Article Google Scholar
• Xing EP, Jordan MI, Russell S et al (2002) Distance metric learning with application to clustering with side-information [C]. In: Advances in neural information processing systems, pp 505–512
• Ying Y, Huang K, Campbell C (2009) Sparse metric learning via smooth optimization [C]. In: Advances in neural information processing systems, pp 2214–2222
• Yu J, Tao D, Li J et al (2014) Semantic preserving distance metric learning and applications [J]. Inf Sci 281:674–686
  Article MathSciNet Google Scholar

Download references

Acknowledgments

This study was funded by National Natural Science Foundation of People’s Republic of China (61173163, 61370200). Huibing Wang, Lin Feng and Yang Liu declare that they have no conflict of interest. This article does not contain any studies with human participants or animals performed by any of the authors.

Author information

Authors and Affiliations

1. School of Computer Science and Technology, Dalian University of Technology, Dalian, 116024, China
   Huibing Wang, Lin Feng & Yang Liu

Authors

1. Huibing Wang
2. Lin Feng
3. Yang Liu

Corresponding author

Correspondence toLin Feng.

Additional information

Communicated by A. Di Nola.

Rights and permissions

About this article

Cite this article

Wang, H., Feng, L. & Liu, Y. Metric learning with geometric mean for similarities measurement.Soft Comput 20, 3969–3979 (2016). https://doi.org/10.1007/s00500-015-1985-x

Download citation

• Published: 21 December 2015
• Issue date: October 2016
• DOI: https://doi.org/10.1007/s00500-015-1985-x

Keywords