Google’s DeepMind AI has pioneered a new technology which allows robots to dream in order to improve their rate of learning.
This follows after the company’s recent announcement of its new lip-reading software using artificial intelligence.
The latest development differs from the previous research, in which the company trained its AI to master a ‘Watch, Listen, Attend and Spell (WLAN) network structure whilst transcribing several hours of BBC footage.
In precise explanation of its latest research, DeepMind explained that the aim of the study “is to recognise phrases and sentences being spoken by a talking face, with or without the audio.
“Unlike previous works that have focused on recognising a limited number of words or phrases, we tackle lip reading as an open-word problem- unconstrained natural language sentences, and in the wild videos.”
Following its announcement earlier this month, DeepMind researchers partnered with Blizzard Entertainment in order to transform one of its games into a learning environment for AI. Its popular video game, StarCraft 2, will be used to teach and test machine agents.
In particular, researchers identified that StarCraft was an “interesting testing environment for current AI research because it provides a useful bridge to the messiness of the real-world.”
The ‘dreams’ discovered by researchers, are to allow the AI to highlight different sections of the gamed that may be particularly challenging and repeat them until mastered.
This technique was found to develop a total of 10 times speed increase in the rate of learning.
In a previous blog post, the company said: “DeepMind is on a scientific mission to push the boundaries of AI, developing programs that can learn to solve any complex problem without needing to be told how.
“Games are the perfect environment in which to do this, allowing us to develop and test smarter, more flexible AI algorithms quickly and efficiently, and also providing instant feedback on how we’re doing through scores.”
The company had previously assisted in pioneering games as a use for AI research environments in a drive for both machine learning and reinforcement learning research, such as Atari games.