We value your privacy

    We use cookies to understand how you interact with our website to improve your experience. By accepting, you agree to our use of these cookies. You can always change your mind later.

    Back to Insights
    Events

    Advancing Science with Data-Efficient AI: CHAIR 2025 Keynote

    IFLAI
    December 19, 2025
    2 min read

    At IFLAI, our core mission has always been to bridge the gap between abstract artificial intelligence and the rigid, physical reality of the natural sciences. Recently, our CTO and Co-founder, Henrik Klein Moberg, had the honor of returning to the Chalmers AI Research Centre (CHAIR) to speak at the Students of AI 2025 event.

    In his presentation, titled "Advancing Science with Data-Efficient AI," Henrik details exactly why standard, brute-force deep learning falls short in scientific applications like early-stage drug discovery, microscopy, and spectroscopy.

    <div className="my-8 aspect-video w-full rounded-xl overflow-hidden shadow-lg border border-border"> <iframe width="100%" height="100%" src="https://www.youtube.com/embed/ZEtBiaobYVQ" title="YouTube video player" frameBorder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" allowFullScreen> </iframe> </div>

    The Data Bottleneck in Science

    A recurring theme in our work is the rejection of the "more data" paradigm when applied to the physical sciences. Generating millions of annotated data points in a wet lab is not just financially prohibitive-it is often physically impossible within a reasonable timeframe.

    During the talk, Henrik breaks down how IFLAI overcomes this bottleneck:

    1. Physics-Informed Architectures: Embedding the laws of physics (like optical transfer functions and geometric equivariance) directly into the neural network architecture.
    2. Data Efficiency: By ensuring the model already "understands" the physics of the environment, it can achieve state-of-the-art accuracy from just a handful of data points, rather than millions.
    3. Real-World Impact: Accelerating pipelines in sectors like Medtech and Advanced Manufacturing, where precision and efficiency are paramount.

    For organizations looking to deploy AI in environments constrained by data availability, the answer isn't larger models. The answer is smarter, domain-aware architectures.

    We’d like to extend our thanks to the Chalmers AI Research Centre for hosting the event and fostering the next generation of AI researchers.