Most artificial intelligences are still built on human labor. Take a look at an AI algorithm and you will find something constructed using data that has been prepared and tagged by an army of human workers.
Now Facebook has shown how some AI algorithms can learn to do useful work with far less human help. The company has built an algorithm that has learned to recognize objects in images with a little help from labels.
“The results are impressive,” said Olga Rusakovski, an assistant professor at Princeton University who specializes in artificial intelligence and computer vision. “Attracting self-supervised learning to work is very challenging, and breakthroughs in this space have important implications for improved visual recognition.”
Rusakovski says it is remarkable that Instagram images are not hand-picked to facilitate self-study.
Facebook research is a benchmark for the AI approach known as “self-directed learning,” says Facebook chief scientist Jan LeKun.
LeCun is a pioneer in the machine learning approach known as deep learning, which involves submitting data to large artificial neural networks. About a decade ago, deep learning emerged as a better way to program machines to do all sorts of useful things, such as image classification and speech recognition.
But LeCun says the conventional approach, which requires “learning” an algorithm by giving it a lot of labeled data, simply won’t scale. “I’ve been advocating for this whole idea of self-directed learning for a long time,” he said. “In the long run, progress in AI will come from programs that just watch videos all day and learn like a baby.”
LeCun says self-directed learning can have many useful applications, such as learning to read medical images without having to label so many scans and X-rays. He says a similar approach is already being used to automatically generate hashtags for Instagram images. And he says Seer technology can be used on Facebook to combine ads with posts or help filter out unwanted content.
Facebook’s research is based on constant progress in changing deep learning algorithms to make them more efficient and effective. Self-directed learning has previously been used to translate text from one language to another, but has been more difficult to apply to images than words. LeCun says the research team has developed a new way for algorithms to learn to recognize images, even when a part of the image has been changed.
Facebook will release some of the technology behind Seer, but not the algorithm itself, as it has been trained using Instagram user data.
Aude Oliva, who heads MIT’s Computational Perception and Knowledge Laboratory, says the approach “will allow us to take on more ambitious visual recognition tasks.” But Oliva says the large size and complexity of state-of-the-art artificial intelligence algorithms like Seer, which can have billions or trillions of neural connections or parameters – much more than a conventional image recognition algorithm with comparable performance – also creates problems. Such algorithms require huge amounts of computing power, making it difficult to supply chips.
Alexei Efros, a professor at UC Berkeley, says the Facebook paper is a good demonstration of an approach he says will be important for AI’s progress – machines to learn on their own using “huge amounts of data”. And as with the greatest advances in artificial intelligence today, he says, it is based on a series of other advances from the same team on Facebook, as well as other research groups in academia and industry.
This story originally appeared on wired.com.