|
Images connect us to information and each other
|
|
Image credits from top left: Mario Testino, Trevor Paglen, Catherine Opie, Edward Burtynsky, Joe Giddens, Ruth Black
|
Images make us feel |
|
My research focuses on the perception of valence. Valence is the rapid, unconscious emotional response to imagery. It can be measured using behavioral experiments and the underlying mechanisms have been uncovered using functional MRI. My most recent work takes valence data and uses it to train convolutional neural networks so that you can predict the engagement with novel images at massive scale.
|
AI to predict images that drive emotion
|
I co-founded Neon Open, an open-source artificial intelligence platform that predicts which images and videos will drive more clicks, likes, and shares. Underlying the platform is a convolutional deep neural network trained on valence data which works by assigning engagement scores to photos and videos. The platform is used by large-scale media publishers, advertisers, researchers, and artists to scan through billions of images to automatically find the most engaging images, thumbnails, and video clips.
|