Google has open-sourced the machine learning technology TensorFlow which is the brain behind their photo search, speech recognition, and the new “Smart Reply” feature in the gmail app. Now it’s available for everyone; students, researchers, hobbyists, hackers, engineers, developers, inventors and innovators.
Google decided to open-source TensorFlow because they believe that machine learning is essential for innovating products and technologies of the future. By sharing it they hope to create a standard for exchanging research ideas and getting machine learning in more products.
Deep learning is a branch of machine learning based on how the brain works. The photos app uses this to analyzing thousands of photos. The AI technology can learn to identify, lets say a chair it has never seen before. The same goes for cats and dogs, doughnuts and clouds. It uses perceptual and language understanding tasks allowing computers to actually see and understand what is in an image or short video clip. So say you are in Japan and trying to read a menu when you don’t read Japanese. You can simply take a picture and it will be able to translate it for you.
Google isn’t the only one who has been working on machine learning though, other companies and researchers have also made huge strides in this area of AI, including Facebook, Microsoft, and Twitter. They too have released open-source code but Google’s move is significant. TensorFlow is depicted by some as the world’s most advanced, because Google.
Artificial Intelligence is playing an increasingly important role in the world’s online services. TensorFlow is great for research but it’s also ready for implementation in real products. Even with TensorFlow in hand, building a deep learning app still requires some heavy engineering. But with the quickly evolving world of Ai this too may change in the years to come.