How Deep Learning Will Change Our World

Comments Off on How Deep Learning Will Change Our World
How Deep Learning Will Change Our World

Deep  learning is a technology that enables machines to identify patterns in vast amounts of “unsupervised” data and to recognize images or speech. Machines are adept at processing supervised data—for example, the data that engineers manually input into a system to enable it to understand spoken commands. The goal of deep learning is to create machines that learn without the need for human input, such as by identifying the words spoken in millions of videos.

According to a report in Re/Code, Andrew Ng, a computer scientist from Stanford who led Google’s deep learning team and is now the chief scientist at Baidu, provides the best definition of deep learning: “It’s a learning technology that works by loosely simulating the brain. Your brain and mine work by having massive amounts of neurons, jam-packed, talking to each other. Deep learning works by having a loose simulation of neurons—hundreds of thousands or millions of neurons—simulating the computer, talking to each other.”1

More precisely, those millions of artificial neurons work in layers. According to MIT Technology Review, the first layer learns simple features that are easy to decipher, such as the edge of an image or a syllable of speech.2 The neurons in this layer achieve this feat by identifying patterns of image pixels or sound waves that are present more often than they would be if they occurred randomly.

After the first layer identifies those features, it passes them on to the next layer, which learns over time to correctly identify features that are more detailed, such as the corner of an image or a word that is formed when syllables are combined. The processing continues through several more layers until the deep learning system identifies the subject of a photo or the sentence formed from a string of words.

While the concept of improving computers by modeling the human brain has been around for decades, it is only now—with advances in computer processing and in the effectiveness of algorithms—that the potential of deep learning has been unlocked. It’s no surprise that Google is at the forefront of the deep learning movement. Google is one of the few companies that can access the computing power to run deep learning experiments.

For example, Google recently harnessed 16,000 computer processors to demonstrate a deep learning system that—after viewing 10 million images from YouTube videos—doubled the accuracy rates of previous image recognition systems in scanning images and classifying the subjects in them, such as cats, yellow flowers, and hockey players...

To continue reading, become a paid subscriber for full access.
Already a Trends Magazine subscriber? Login for full access now.

Subscribe for as low as $195/year

  • Get 12 months of Trends that will impact your business and your life
  • Gain access to the entire Trends Research Library
  • Optional Trends monthly CDs in addition to your On-Line access
  • Receive our exclusive "Trends Investor Forecast 2015" as a free online gift
  • If you do not like what you see, you can cancel anytime and receive a 100% full refund