“We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g., freckles, hair), and it enables intuitive, scale-specific control of the synthesis. The new generator improves the state-of-the-art in terms of traditional distribution quality metrics, leads to demonstrably better interpolation properties, and also better disentangles the latent factors of variation. To quantify interpolation quality and disentanglement, we propose two new, automated methods that are applicable to any generator architecture. Finally, we introduce a new, highly varied and high-quality dataset of human faces.”
“In a world where surveillance technology is being deployed everywhere from airports and stadiums to public schools and hotels and raising a plethora of privacy concerns, it’s perhaps inevitable that farms on land and at sea would find ways to exploit it to improve productivity. Just this year, American agribusiness giant Cargill Inc. said it was working with an Irish tech start-up on a facial-recognition system to monitor cows so farmers can adjust feeding regimens to enhance milk production. Scanners will allow them to track food and water intake and even detect when females are having fertile days. Salmon farming may be next in line. As fish vies with beef and chicken as the global protein food of choice, exporters like Norway, the world’s biggest producer of the pinkish-orange fish, have become the focal point for radical marine-farming methods designed to help the $232 billion aquaculture industry feed the world.”
“Creating slow-motion footage is all about capturing a large number of frames per second. If you don’t record enough, it becomes choppy and unwatchable as soon as you slow down your video — unless, that is, you use artificial intelligence to imagine the extra frames”.
«We present a method to create universal, robust, targeted adversarial image patches in the real world. The patches are universal because they can be used to attack any scene, robust because they work under a wide variety of transformations, and targeted because they can cause a classifier to output any target class. These adversarial patches can be printed, added to any scene, photographed, and presented to image classifiers; even when the patches are small, they cause the classifiers to ignore the other items in the scene and report a chosen target class».
«Arsenal’s smart assistant AI suggests settings based on your subject and environment. It uses an advanced neural network to pick the optimal settings for any scene (using similar algorithms to those in self driving cars). Like any good assistant, it then lets you control the final shot. Here’s how it works…»
«Unsupervised image-to-image translation aims at learning a joint distribution of images in different domains by using images from the marginal distributions in individual domains. Since there exists an infinite set of joint distributions that can arrive the given marginal distributions, one could infer nothing about the joint distribution from the marginal distributions without additional assumptions».