Learning using Privileged Information

For 2,000 years, we believed logic was the only instrument for solving intellectual problems.
Now, our analysis of machine learning is showing us that to address truly complex problems, we need images, poetry, and metaphors as well.” Vladimir Vapnik.

Learning using privileged information (LUPI) was first introduced by Vapnik and Vashist1 as a framework for learning with additional information at training time that is not available at test time. LUPI framework seeks to devise learning methods that can utilize this additional information with an emphasis that the privileged information is not required or more importantly not needed to be inferred at test time. Thus, this information cannot be simply used as an input to the learner.

1 V. Vapnik and A. Vashist. A new learning paradigm: Learning using privileged information. Neural Networks, pp. 544--557, 2009.


  • We are organising the 1st Workshop on Human is More Than a Labeler (co-located with IJCAI 2016 in New York City). One of our invited speakers is Prof. Vladimir Vapnik from Facebook AI Research, who introduced learning using privileged information. For more information: http://smileclinic.alwaysdata.net/ijcai16workshop/.
  • We have 2 papers accepted at CVPR 2016.

  • Our Contributions

    We show that the LUPI framework is applicable to several scenarios that have been studied in computer vision before: attributes, bounding boxes and image tags. By exploiting the insight that privileged information allows us to distinguish between easy and hard examples in the training set, we have introduced RankTransfer, MarginTransfer, and PrivilegedNoise methods. Recently, we are exploring the usage of privileged information for incorporating training label annotations confidence and for performing cross dataset learning.

    • V. Sharmanska and N. Quadrianto. Learning from the Mistakes of Others: Matching Errors in Cross Dataset Learning. Computer Vision and Pattern Recognition (CVPR (spotlight)), 2016.
    • V. Sharmanska, D. Hernández-Lobato, J. M. Hernández-Lobato, and N. Quadrianto. Ambiguity Helps: Classification with Disagreements in Crowdsourced Annotations. Computer Vision and Pattern Recognition (CVPR), 2016.
    • V. Sharmanska and N. Quadrianto. In the Era of Deep Convolutional Features: Are Attributes still Useful Privileged Data? Springer, Rogerio Feris, Devi Parikh, Christoph H. Lampert (Eds.), 2016.
    • D. Hernández-Lobato, V. Sharmanska, K. Kersting, C. H. Lampert, and N. Quadrianto. Mind the Nuisance: Gaussian Process Classification using Privileged Noise. Neural Information Processing Systems (NIPS), 2014.
    • V. Sharmanska, N. Quadrianto, and C.H. Lampert. Learning to Transfer Privileged Information. arXiv:1410.0389, 2014.
    • V. Sharmanska, N. Quadrianto, and C.H. Lampert. Learning to Rank Using Privileged Information. International Conference on Computer Vision (ICCV), 2013.

    Research Codes

    Research Team @ University of Sussex