Organizing Committee

  • Andrey Malinin
    Andrey Malinin
  • Vatsal Raina
    Vatsal Raina
  • Mark Gales
    Mark Gales
  • Mara Graziani
    Mara Graziani
  • Meritxell Bach Cuadra
    Meritxell Bach Cuadra
  • Nataliia Molchanova
    Nataliia Molchanova
  • Cristina Granziera
    Cristina Granziera
  • Elena Volf
    Elena Volf
  • Nikolay Kartashev
    Nikolay Kartashev
  • Konstantinos Kyriakopoulos
    Konstantinos Kyriakopoulos
  • Efi Tsompopoulou
    Efi Tsompopoulou
  • Antonis Nikitakis
    Antonis Nikitakis
  • Andreas Athanasopoulos
    Andreas Athanasopoulos

Who we are

The Shifts Project’s mission is to encourage research collaboration into the application and evaluation of deep learning models under conditions of in-the-wild distributional shift. Our vision is that increased research interest in this area will enable the application of deep learning to a more diverse range of real-life applications, lead to the development of more robust and higher performing models, create tools for the utility, safety and suitability for downstream optimisation tasks of deep learning models to be more fairly evaluated, and unlock greater understanding of the properties and robustness of neural networks. To this end, the independent Organising Committee has been formed in 2022 for the purposes of:

  • Building partnerships between companies, institutions and researchers to encourage research collaboration.
  • Making available public datasets from diverse in-the-wild applications of deep learning to the research community.
  • Establishing and raising awareness of benchmarks that allow the robustness of models to distributional shift to be fairly evaluated.
  • Organising challenges, workshops, talks and other events to encourage research into distributional shifts.

The 2021 Shifts Challenge was co-organised by the University of Cambridge, the University of Oxford and Yandex Research, with challenge data provided by Yandex Self-driving Car. Yandex is not a data provider of the 2022 Shifts Challenge.