Meaningful Signals Within Deep Learning Models for Biology - Primer and Meeting

Explore cutting-edge research on meaningful signals within deep learning models for biological applications in this comprehensive conference talk from the Broad Institute. Delve into three key areas: the development of meaningful pretrained models for biology, the curation of pre-training data aligned with downstream tasks, and techniques for disentangling meaningful signals from experimental noise. Learn about the challenges and potential solutions in applying deep learning to biomedical imaging data, including the creation of CytoImageNet, a large-scale dataset of microscopy images. Discover how batch effects normalization (BEN) can significantly improve both supervised and unsupervised learning in high-throughput microscopy experiments. Gain valuable insights into the latest advancements in computational biology and their implications for future research and applications.

Primer: Towards Meaningful Pretrained Models for Biology
Meeting Part 1: Meaningful choice/curation of pre-training data in alignment with a downstream task
Meeting Part 2: Disentangling Meaningful Signal from Experimental Noise within Deep Learning Models