
Juan C. Caicedo
@jccaicedo
Schmidt Fellow at the Broad Institute of MIT and Harvard
ID: 69134120
26-08-2009 23:37:05
447 Tweet
358 Followers
146 Following

We evaluated a couple of deep learning architectures vs. classical image processing on 20,000 manually annotated nuclei! Paper from Juan C. Caicedo in BioRxiv! biorxiv.org/content/early/…





Using auxiliary tasks is a common way to speed up learning, though it's not clear a priori if/when a given aux task will help the main task. We propose a simple method that adapts aux losses using cosine similarity of gradients: arxiv.org/abs/1812.02224 Balaji Lakshminarayanan @rpascanu Sid Jayakumar




Want to devote your expertise in machine learning to accelerate the pace at which new medicines are found? Join Anne Carpenter, PhD and me at the Broad Institute to glean insights from biological images! Learn more about this postdoctoral position at broad.io/mlcbpostdoc.






Neat trick to train deep learning models without conventional biological ground truth: Weakly supervised learning of single-cell feature embeddings. Postdoc Juan C. Caicedo's paper coming out at CVPR: biorxiv.org/content/early/…


An overview of the computational methods we use for image-based profiling. Together with @snhantau Jane Hung, and Mohammad Hossein Rohban youtu.be/QSfKbYLLkoc