Big tech companies like Amazon, Facebook, and now Apple are publishing their research in an information-sharing effort to advance budding artificial intelligence technology (read: hasten the machine takeover).
Apple’s first contribution is a paper, published December 22, titled “Learning from Simulated and Unsupervised Images through Adversarial Training.” The authors listed are Ashish Shrivastava, Tomas Pfister, Oncel Tuzel, Josh Susskind, Wenda Wang, and Russ Webb. The research concerns teaching an algorithm to recognize random images by training it with computer-generated pictures that are fully labeled. Apple already uses “computer vision” AI to help recognize images in its Photos app, but the new research could help Apple recognize a far wider array of images with more accuracy and detail.
On December 6, Apple’s head of machine learning research, Russ Salakhutdinov, said at an AI conference in Spain that Apple would begin publishing its AI research. This was a surprising, and hopeful, move for the notoriously secret Apple.