Deep learning can be applied to tasks that involve a breadth of data types. I talk with Kyle Swanson, currently a Marshall Scholar at Cambridge University and previously at MIT, about his projects with Regina Barzilay on predicting breast cancer from mammograms, with performance on par with radiologists, and on designing antibiotics in a high-throughput manner using a new class of neural networks, identifying a compound that traditional chemists may have overlooked.

Check out the glossary of terms, definitions, and resources (and get a sneak peak of the future conversations lined up!) here: bit.ly/datapulse-glossary