Optimizing Ranking Measures for Compact Binary Code Learning

  • Guosheng Lin
  • Chunhua Shen
  • Jianxin Wu
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8691)

Abstract

Hashing has proven a valuable tool for large-scale information retrieval. Despite much success, existing hashing methods optimize over simple objectives such as the reconstruction error or graph Laplacian related loss functions, instead of the performance evaluation criteria of interest—multivariate performance measures such as the AUC and NDCG. Here we present a general framework (termed StructHash) that allows one to directly optimize multivariate performance measures. The resulting optimization problem can involve exponentially or infinitely many variables and constraints, which is more challenging than standard structured output learning. To solve the StructHash optimization problem, we use a combination of column generation and cutting-plane techniques. We demonstrate the generality of StructHash by applying it to ranking prediction and image retrieval, and show that it outperforms a few state-of-the-art hashing methods.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Chakrabarti, S., Khanna, R., Sawant, U., Bhattacharyya, C.: Structured learning for non-smooth ranking losses. In: Proc. ACM Knowledge Discovery & Data Mining (2008)Google Scholar
  2. 2.
    Demiriz, A., Bennett, K.P., Shawe-Taylor, J.: Linear programming boosting via column generation. Mach. Learn. (2002)Google Scholar
  3. 3.
    Gionis, A., Indyk, P., Motwani, R.: Similarity search in high dimensions via hashing. In: Proc. Int. Conf. Very Large Data Bases (1999)Google Scholar
  4. 4.
    Gong, Y., Lazebnik, S., Gordo, A., Perronnin, F.: Iterative quantization: A procrustean approach to learning binary codes for large-scale image retrieval. IEEE Trans. Patt. Anal. & Mach. Intelli. (2012)Google Scholar
  5. 5.
    Heo, J., Lee, Y., He, J., Chang, S., Yoon, S.: Spherical hashing. In: Proc. Int. Conf. Comp. Vis. & Patt. Recogn. (2012)Google Scholar
  6. 6.
    Järvelin, K., Kekäläinen, J.: IR evaluation methods for retrieving highly relevant documents. In: Proc. ACM Conf. SIGIR (2000)Google Scholar
  7. 7.
    Joachims, T.: A support vector method for multivariate performance measures. In: Proc. Int. Conf. Mach. Learn. (2005)Google Scholar
  8. 8.
    Joachims, T.: Training linear SVMs in linear time. In: Proc. ACM Knowledge Discovery & Data Mining (2006)Google Scholar
  9. 9.
    Kelley Jr., J.E.: The cutting-plane method for solving convex programs. J. Society for Industrial & Applied Math. (1960)Google Scholar
  10. 10.
    Kulis, B., Darrell, T.: Learning to hash with binary reconstructive embeddings. Proc. Adv. Neural Info. Process. Syst. (2009)Google Scholar
  11. 11.
    Kulis, B., Grauman, K.: Kernelized locality-sensitive hashing. IEEE Trans. Patt. Anal. & Mach. Intelli. (2012)Google Scholar
  12. 12.
    Li, X., Lin, G., Shen, C., van den Hengel, A., Dick, A.: Learning hash functions using column generation. In: Proc. Int. Conf. Mach. Learn. (2013)Google Scholar
  13. 13.
    Lin, G., Shen, C., Shi, Q., van den Hengel, A., Suter, D.: Fast supervised hashing with decision trees for high-dimensional data. In: Proc. Int. Conf. Comp. Vis. & Patt. Recogn. Columbus, Ohio, USA (2014), https://bitbucket.org/chhshen/fasthash/src
  14. 14.
    Lin, G., Shen, C., Suter, D., van den Hengel, A.: A general two-step approach to learning-based hashing. In: Proc. Int. Conf. Comp. Vis., Sydney, Australia (2013)Google Scholar
  15. 15.
    Liu, W., Wang, J., Ji, R., Jiang, Y., Chang, S.: Supervised hashing with kernels. In: Proc. Int. Conf. Comp. Vis. & Patt. Recogn. (2012)Google Scholar
  16. 16.
    Liu, W., Wang, J., Kumar, S., Chang, S.F.: Hashing with graphs. In: Proc. Int. Conf. Mach. Learn. (2011)Google Scholar
  17. 17.
    McFee, B., Lanckriet, G.: Metric learning to rank. In: Proc. Int. Conf. Mach. Learn. (2010)Google Scholar
  18. 18.
    Shen, C., Li, H.: On the dual formulation of boosting algorithms. IEEE Trans. Patt. Anal. & Mach. Intelli. (2010)Google Scholar
  19. 19.
    Shen, C., Lin, G., van den Hengel, A.: StructBoost: Boosting methods for predicting structured output variables. IEEE Trans. Patt. Anal. & Mach. Intelli. (2014)Google Scholar
  20. 20.
    Shen, F., Shen, C., Shi, Q., van den Hengel, A., Tang, Z.: Inductive hashing on manifolds. In: Proc. Int. Conf. Comp. Vis. & Patt. Recogn., Oregon, USA (2013)Google Scholar
  21. 21.
    Tsochantaridis, I., Hofmann, T., Joachims, T., Altun, Y.: Support vector machine learning for interdependent and structured output spaces. In: Proc. Int. Conf. Mach. Learn. (2004)Google Scholar
  22. 22.
    Wang, J., Kumar, S., Chang, S.: Semi-supervised hashing for large scale search. IEEE Trans. Patt. Anal. & Mach. Intelli. (2012)Google Scholar
  23. 23.
    Weiss, Y., Fergus, R., Torralba, A.: Multidimensional spectral hashing. In: Proc. Eur. Conf. Comp. Vis. (2012)Google Scholar
  24. 24.
    Weiss, Y., Torralba, A., Fergus, R.: Spectral hashing. In: Proc. Adv. Neural Info. Process. Syst. (2008)Google Scholar
  25. 25.
    Yue, Y., Finley, T., Radlinski, F., Joachims, T.: A support vector method for optimizing average precision. In: Proc. ACM Conf. SIGIR (2007)Google Scholar
  26. 26.
    Zhang, D., Wang, J., Cai, D., Lu, J.: Extensions to self-taught hashing: Kernelisation and supervision. In: Proc. ACM Conf. SIGIR Workshop (2010)Google Scholar
  27. 27.
    Zhu, C., Byrd, R.H., Lu, P., Nocedal, J.: Algorithm 778: L-BFGS-B: Fortran subroutines for large-scale bound-constrained optimization. ACM T. Math. Softw. (1997)Google Scholar

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  • Guosheng Lin
    • 1
  • Chunhua Shen
    • 1
  • Jianxin Wu
    • 2
  1. 1.University of AdelaideAustralia
  2. 2.Nanjing UniversityChina

Personalised recommendations