Biotech Breakfast: Exploring Deep Learning in Biomedical Science

​Deep learning has achieved significant breakthroughs in natural language processing, sparking great interest in applying these methods to biomedical data.

​However, this transition presents unique technical challenges, opportunities, and risks. In this talk, we will explore biomedical foundation models, including protein language models and AlphaFold, explaining the core technologies behind them in an accessible manner.

​We will also discuss the transformative potential of these models in biology and medicine, while addressing ethical concerns, including the risk of misuse in developing harmful biological agents.

​

​Konstantin Dobler is a PhD student in Machine Learning at the Hasso Plattner Institute, University of Potsdam, Germany, under the supervision of Gerard de Melo. He is also a member of the ELLIS network and co-advised by Desmond Elliott from the University of Copenhagen.

​Currently, he is completing a research internship at InstaDeep, where he focuses on protein design and multimodal modeling. His research interests include transfer learning  of pretrained models, representational alignment across data modalities, and improving the computational efficiency of deep learning systems.