Leading artificial intelligence experts are investigating ways to commercialize a rapidly emerging sub-field of research known as “deep learning”. It was revived in 2006 by renowned scientist Geoffrey E. Hinton. This month, a research team under Hinton’s tutelage won a prize sponsored by Merck to design software to uncover molecules that are most likely to be good candidates for new drugs.
The win was a particularly impressive feat given that the team entered at the last minute, and were working with relatively small data-sets. Click here to read more about “how they did it.”
In a story originally reported by the New York Times, even skeptical scientists admit this is a significant advancement. “The kind of jump we are seeing in the accuracy of these systems is very rare indeed,” NYU computer scientist Yann leCun told the Times.
Silicon Valley’s technology companies have utilized artificial intelligence technology for several years. Deep learning is yielding new discoveries in fields like speech recognition and computer vision. It is already used in Apple’s Siri virtual personal assistant and Google Street View.
This is just the beginning. Today’s image recognition systems do not use human-familiar concepts like ‘mouth, or ‘eyes’ but statistical properties derived from the image. Deep learning is based on learning several levels of representations, and higher-level concepts are defined from lower-level ones. “The innovation of deep learning is that it not only arranges these properties into hierarchies, but it works out how many levels of hierarchy best fit the data,” explain Tom Stafford and Matt Webb, neuroscience researchers from Mind Hacks.
In the future, New York Times’ science writer John Markoff posits that deep learning will make surveillance technologies cheaper and more accessible; it may also pave the way for self-driving cars and robots that can replace human workers.
Comments are disabled on this post