Don't miss the latest stories
Google Teaches Its Artificial Intelligence A Brand New Trick: Taking Naps
By Mikelle Leow, 11 Aug 2017
Composite image by DesignTAXI. Google logo via Wikimedia Commons via Shutterstock
A month after Google’s ‘DeepMind’ AI strangely taught itself to walk without even being told to, the company has announced another milestone: its newest trick is taking a nap.
Google understands why you might believe that a napping machine ruins the point of building a bot altogether. The tech giant reasons, however, that it’s teaching ‘DeepMind’ to become more human—and humans do need sleep.
In a blog post, DeepMind researchers explain, “At first glance, it might seem counter-intuitive to build an artificial agent that needs to ‘sleep’—after all, they are supposed to grind away at a computational problem long after their programmers have gone to bed.”
“But this principle was a key part of our deep-Q network (DQN), an algorithm that learns to master a diverse range of Atari 2600 games to superhuman level with only the raw pixels and score as inputs. DQN mimics ‘experience replay’, by storing a subset of training data that it reviews ‘offline’, allowing it to learn anew from successes or failures that occurred in the past.”
Simply put, researchers are training ‘DeepMind’ to learn and figure things out by itself. In teaching the AI to nap, the machine might be using its full capacity when at work, but it can save excess information to dream about later, when it’s offline. Talk about sleeping on it.
[via The Next Web, cover image via various sources]
More related news
Also check out these recent news