Supervised Asymmetric Metric Extraction: An Approach to Combine Distances

  • Archil Maysuradze
  • B. H. Shekar
  • Mikhail Suvorov
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10682)

Abstract

We propose a novel supervised distance metric extraction technique. Given several original metrics and a finite set of labeled objects, the problem is to produce a new metric which better agrees with the labels of the training objects. The problem may be seen as the best single metric extraction from a metric-based description. Feature-based object descriptions are not used even implicitly. Unlike many metric approaches, we treat intraclass and interclass distances differently. The metric extraction problem is reduced to a linear programming problem that makes it possible to use effective optimization techniques. It is proved that an admissible solution always exists and hence there is no need to introduce any soft-constraint extension and the number of variables remains small. Thus, the computational complexity depends mainly on the original metric calculation. The method is empirically tested on biometric data where all the original and derived metrics are calculated in real time.

Keywords

Multiple distance metrics Similarity measures Metric-based descriptions Dimensionality reduction Combined classifier Constrained optimization 

Notes

Acknowledgments

This work was partially supported by Lomonosov Moscow State University research project “Algebraic, logical and statistical machine learning methods and their application in applied data analysis”, RFBR projects No 15-07-09214, 16-01-00196, 16-57-45054 and DST-RFBR Grant No. INT/RUS/RFBR/P-248.

References

  1. 1.
    Maysuradze, A.I., Suvorov, M.A.: Aggregation of multiple metric descriptions from distances between unlabeled objects. J. Comput. Math. Math. Phys. 57(2), 350–361 (2017)MathSciNetCrossRefMATHGoogle Scholar
  2. 2.
    Bellet A., Habrard, A., Sebban, M.: A Survey on Metric Learning for Feature Vectors and Structured Data CoRR abs/1306.6709 (2013)Google Scholar
  3. 3.
    Weinberger, K.Q., Saul, L.K.: Distance metric learning for large margin nearest neighbor classification. J. Mach. Learn. Res. 10, 207–244 (2009)MATHGoogle Scholar
  4. 4.
    Wang, Y., Lin, X., Zhang, Q.: Towards metric fusion on multi-view data: a cross-view based graph random walk approach. In: Proceedings of the 22nd ACM International Conference on Information & Knowledge Management (CIKM 2013), pp. 805–810 (2013)Google Scholar
  5. 5.
    Davis, J.V., Kulis, B., Jain, P., Sra, S., Dhillon, I.S.: Information-theoretic metric learning. In: Proceedings of the 24th international conference on Machine learning, pp. 209–216. ACM (2007)Google Scholar
  6. 6.
    Maysuradze, A.I.: On optimal decompositions of finite metric configurations in pattern recognition problems. J. Comput. Math. Math. Phys. 44(9), 1615–1624 (2004)MathSciNetGoogle Scholar
  7. 7.
    Maysuradze, A.I.: Homogeneous and rank bases in spaces of metric configurations. J. Comput. Math. Math. Phys. 46(2), 330–344 (2006)MathSciNetCrossRefGoogle Scholar
  8. 8.
    Mestetskiy, L.: Shape comparison of flexible objects: similarity of palm silhouettes. In: Proceedings of the 2nd International Conference on Computer Vision Theory and Applications (VISAPP), pp. 390–393 (2007)Google Scholar
  9. 9.
    Mestetskiy, L., Bakina, I., Kurakin, A.: Hand geometry analysis by continuous skeletons. In: Kamel, M., Campilho, A. (eds.) ICIAR 2011. LNCS, vol. 6754, pp. 130–139. Springer, Heidelberg (2011).  https://doi.org/10.1007/978-3-642-21596-4_14 CrossRefGoogle Scholar
  10. 10.
    Bakina, I., Mestetskiy, L.: Hand shape recognition from natural hand position. In: Proceedings of the IEEE International Conference on Hand-Based Biometrics. The Hong Kong Polytechnic University, Hong Kong, pp. 170–175 (2011)Google Scholar
  11. 11.
    Chernyshov, V., Mestetskiy, L.: Mobile computer vision system for hand-based identification. In: Pattern Recognition and Image Analysis, vol. 25, no. 2, pp. 209–214. Allen Press Inc., Cambridge (2015)Google Scholar
  12. 12.
    Suryanto, C.H., Hino, H., Fukui, K.: Combination of multiple distance measures for protein fold classification. In: 2nd IAPR Asian Conference on Pattern Recognition (ACPR), pp. 440–445. IEEE (2013)Google Scholar
  13. 13.
    Jang, D., Jang, S.-J., Lim, T.-B.: Distance combination for content identification system. In: 2013 1st International Conference on Communications, Signal Processing, and their Applications (ICCSPA), pp. 1–6 (2013)Google Scholar
  14. 14.
    Bellet, A., Habrard, A., Sebban, M.: Metric Learning, Synthesis Lectures on Artificial Intelligence and Machine Learning. Morgan & Claypool Publishers, San Rafael (2015)MATHGoogle Scholar
  15. 15.
    Kulis, B.: Metric learning: a survey. In: Foundations and Trends in Machine Learning, vol. 5, no. 4, pp. 287–364 (2013)Google Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  • Archil Maysuradze
    • 1
  • B. H. Shekar
    • 2
  • Mikhail Suvorov
    • 3
  1. 1.Lomonosov Moscow State UniversityMoscowRussia
  2. 2.Mangalore UniversityMangaloreIndia
  3. 3.CitySoft LLCMoscowRussia

Personalised recommendations