Convolutional networks beyond the Euclidean space

  • * Register (or log in) to the Neural Network to add this session to your agenda or watch the replay

  • Date
    12 November 2025
    Timeframe
    17:00 - 18:00 CET Geneva
    Duration
    60 minutes

    Convolutional networks have revolutionized image analysis and data science. They are among the main driving forces behind the current AI revolution. This presentation outlines key developments, beginning with the initial breakthrough in image classification (Krizhevsky et al., NeurIPS 2012), and then explores efforts to extend the capabilities of deep networks to broader challenges in image analysis, shape analysis, medicine, and the life sciences. Examples include optical flow estimation (Dosovitskiy et al., ICCV 2015), the acceleration of diffusion tensor imaging (Golkov et al. 2016), tuberculosis screening from chest x-rays (Golkov et al., Sci. Reports 2019), and protein structure prediction (Golkov et al., NeurIPS 2016). It also includes a brief overview of convolutional networks for graph-valued data and introduces a recent solution that extends these networks to directed graphs using holomorphic functional calculus (Koke & Cremers, ICLR 2024).

     

    Session Objectives:

    By the end of the session, participants will be able to:

    • Explain the fundamental principles and architecture of convolutional networks.
    • Apply convolutional networks to non-Euclidean data structures.
    • Explore real-world applications of convolutional networks in medical imaging and the life sciences.
  • Discovery badge available
    Showcase it on your LinkedIn profile or your professional portfolio
    Verified badge is available for CHF 10
    Learn more
    Share this session with your network
    Discover more from programme stream

    Generative AI and magical thinking

    12 January 2026 - 16:00 to 17:00

    Are you sure you want to remove this speaker?