Naik, Nikhil, Scott Duke Kominers, Ramesh Raskar, Edward L. Glaeser, César A. Hidalgo. 2017. “Computer vision uncovers predictors of physical urban change.” PNAS 114(29):7571–7576.
Janelle Shane reached out to Twitter asking people to share photos of 'sheep in unusual places' to test her hypothesis that a neural net algorithm was only tagging sheep based on its background - like grass, leaves, etc and hadn't actually learned to recognize sheep themselves. This turns out to be true! For instance, a little girl holding a sheep was tagged as a puppy because the training data had pictures of children with pets. It's definitely v interesting to explore other correlated characteristics of images that lead deep learning algorithms to recognize them as one thing instead of another.
I want to share this wonderful thread from Twitter this morning that is very relevant to this paper - https://twitter.com/JanelleCShane/status/969239712190746624
Janelle Shane reached out to Twitter asking people to share photos of 'sheep in unusual places' to test her hypothesis that a neural net algorithm was only tagging sheep based on its background - like grass, leaves, etc and hadn't actually learned to recognize sheep themselves. This turns out to be true! For instance, a little girl holding a sheep was tagged as a puppy because the training data had pictures of children with pets. It's definitely v interesting to explore other correlated characteristics of images that lead deep learning algorithms to recognize them as one thing instead of another.