Scientists at Google have begun developing algorithms which will allow the tech giant’s ‘DeepMind’ system to see and understand spatial environments.
The research takes inspiration from the way children make sense of the world; interpreting surroundings without bias and labels.
Speaking on the developments, Google DeepMind research scientist S. M. Ali Eslami said, “Think of how an infant might interact with a laptop, to begin with.
“First it might think, ‘that is just a collection of boxes’ … but over time, if it sees the laptop enough … it will figure out that actually this thing, this configuration of boxes, appears to happen quite a lot, and maybe it’s a concept of its own,” he added.
The mechanics behind this lacking bias has been dubbed the Generative Query Network (GQN) and used what’s known as unsupervised learning.
According to a recent article in Science titled ‘Neural scene representation and rendering’, Google DeepMind research scientists have ” developed an artificial vision system, dubbed the Generative Query Network (GQN), that has no need for such labelled data”.
“Instead, the GQN first uses images taken from different viewpoints and creates an abstract description of the scene, learning its essentials.
“Next, on the basis of this representation, the network predicts what the scene would look like from a new, arbitrary viewpoint,” it adds.
Check out the full article here.