Neural networks are powerful prediction tools, but they are too large and inefficient for device-centered applications, and require expensive cloud infrastructure. MIT CSAIL student Lucas Liebenwein hopes to change that. He and his team are working on reducing large networks into smaller architecture, enabling them to be deployed onto small devices like phones and robots. Once deployed, the networks can make faster predictions while maintaining accuracy. Read more about Lucas Liebenwein in his Alliances Spotlight. To view the transcript, click here.