DNNresearch Inc, a 2012 startup from the University of Toronto, has confirmed that Google bought it to improve context-based recognition of voices and imagery. DNNresearch's expertise is in neural networks and computerized learning, based on the research of professor Geoffrey Hinton—a chap who is now going to wear two hats, one for the university and one for Google.
Voice and image recognition are hard enough tasks when computers are asked to make direct recognition decisions like "did the user just say 'red'?" But as anyone who as interacted with Siri, a voice dictation service like those provided by Dragon or even an object-based recognition system like Google Goggles, computers find it very hard to perform context-based recognition such as deciding between "red" and "read," or if the blurry object seen in a video is a indeed a human face. Google's most likely going to leverage DNN's research into services like Google Now, that already try to deliver data to a user based on contextual pointers, its Glass systems, which already include voice recognition, its purported Siri rival, and other efforts to categorize the world.
[Image: Flickr user eagleapex]