A common approach in drug and herbicide discovery consists of testing thousands of chemicals with the help of robotic systems. These large datasets are particularly suitable for training data-greedy deep learning models. However, as many projects are explorative in nature, they lack labels for training supervised models. Therefore, we often use self-supervised learning (SSL) methods to analyze these large, unlabeled datasets. SSL methods use pretext tasks to extract meaningful feature representations of data which can be utilized in downstream data analysis. This talk will show examples of how we leverage SSL methods to accelerate drug and herbicide research at Bayer AG, with focus on computer vision applications.
Paula is a data scientist from Bayer AG in Berlin. She obtained a PhD in biology from the German Cancer Research Center DKFZ, a MSc in Applied Mathematics from Eindhoven University of Technology, and a BSc in Biological Engineering from the National University of Colombia in Medellin. In 2017, she joined Bayer as a postdoctoral fellow, where she developed deep learning methods for phenotypic profiling in plant sciences and cellular images. Since 2019, she works at the Machine Learning Research group from Bayer R&D, where she focuses on image analysis applications to drug discovery.
