How might deep learning be able to produce emotion-inducing content at scale in the future? What content do individuals find more or less emotionally intense? Our team is trying to find out. We’ve combined a crowdsourced image-rating platform with deep learning style transfer tools. Our goal is to collect as many user-rated images as possible and to evaluate the ability of these images to produce changes in emotional states. If interested, help us out (disclaimer, some of these images are pretty scary)! This project is in collaboration with Pinar Yanardag, Manuel Cebrian, and Iyad Rahwan.