AI News: Apple Finally Publishes First Artificial Intelligence Research Paper
After being criticized for keeping their artificial intelligence (AI) research private, Apple has finally published its first paper on artificial intelligence.
Business Insider reported that the iPhone maker published its first AI academic paper titled Learning From Simulated and Unsupervised Images Through Adversarial Training, which focuses on a technique where a learning algorithm could be able to "see" and recognize images.
Published through the Cornell University Library, the study's lead author, Ashish Shrivastava, along with his team Russ Webb, Tomas Pfister, Wenda Wang, Josh Susskind and Oncel Tuzel reported some of Apple's procedures in developing AI.
According to the team, the use of synthetic or animated images could be more efficient in training AI models, which are also known as neural networks to identify rather than using real-world images. However, they are "often not realistic enough, leading the network to learn details only present in synthetic images," which makes these networks "fail to generalize well on real images."
To solve this problem, the researchers pitched the technique "Simulated+Unsupervised learning," which is a combination of annotated synthetic images and data of unlabeled real images.
"With recent progress in graphics, it has become more tractable to train models on synthetic images, potentially avoiding the need for expensive annotations," the researchers wrote, as quoted from the study's abstract. "However, learning from synthetic images may not achieve the desired performance due to a gap between synthetic and real image distributions."
"We develop a method for S+U learning, which we term SimGAN, that refines synthetic images from a simulator using a neural network which we call the 'refiner network,'" the researchers explained. "To add realism - the first requirement of an S+U learning algorithm - we train our refiner network using an adversarial loss."
This puts Apple back in the AI game along with Google, Amazon, Microsoft, Facebook Artificial Intelligence Research (FAIR) and Elon Musk's OpenAI.