With “Project Adam”, Microsoft Claims It’s Bested Google’s AI
With Project Adam, Microsoft has thrown down the artificial intelligence gauntlet. The company boasts that it's new deep-learning system is better in terms of speed, efficiency, and accuracy than Google's best attempts.
With Project Adam, Microsoft has thrown down the artificial intelligence gauntlet. The company boasts that it’s new deep-learning system is better in terms of speed, efficiency, and accuracy than Google’s best attempts at artificial intelligence to date. The key difference is a system Microsoft uses called HOGWILD!, developed at the University of Wisconsin, which allows computers to work in an asynchronous way with each other. “Although neural nets are extremely dense and the risk of data collision is high, this approach works because the collisions tend to result in the same calculation that would have been reached if the system had carefully avoided any collisions.”
What’s the Big Idea?
So what can Project Adam really do? For now, it can identify the content of photos very quickly and with a very high accuracy, differentiating different breeds of dog and even their sub-breeds, for example. That might sound a little gimmicky, but if the results can be scaled and applied to other industries, the effects could become enormously beneficial. “Imagine if you could photograph the food you’re eating and could instantly receive accurate nutrition information to make smarter choices,” says Microsoft. “Imagine if we could identify serious disease well before it spreads based on a photo and some basic lab results.”