What is TensorFlow by Google Brain and How it Works

Everyone who has tried Google Photos would concur that this complimentary photo storage and direction service from Google is smart. It packs in various smart features similar advanced search, ability to categorize your pictures by locations and dates, automatically create albums and videos based on similarities, and walk you lot down the retentiveness lane by showing you lot photos of the same day several years ago. There are many things Google Photos can do that several years ago would be automobile-ly impossible. Google Photos is one of the many "smart" services from Google that uses a machine learning engineering science called TensorFlow. The word learning indicates that the engineering science volition get smarter by fourth dimension to the point that our electric current noesis cannot imagine. But what is TensorFlow? How tin a car learn? What can you lot do with it? Let's discover out.

What is TensorFlow?

TensorFlow is Google's open up-source and powerful artificial intelligence software, which powers many services and initiatives from Google. It is the second generation of a system for big-scale machine learning implementations, congenital by the Google Brain team. This library of algorithm succeeds DistBelief – the first generation.

The engineering science represents computation as stateful data flow graphs. What makes TensorFlow unique is its power to model computations on a vast range of hardware, from consumer-level mobile devices to world-course multi-GPU servers. It can run on different GPUs and CPUs and promises the scalability of machine learning among the various devices and gadgets without having to alter a pregnant amount of code.

TensorFlow originated from Google's demand to instruct a reckoner system to mimic how a human being encephalon works in learning and reasoning. The system, known as neural networks, should be able to perform on multidimensional data arrays referred every bit "tensors." The end goal is to train the neural networks to discover and decipher patterns and correlations.

Tensor -bb- tensor flowing

In November 2015, Google made this technology open up-source and allowed it to be adopted into all kinds of products and researches. Anybody, including researchers, engineers, and hobbyists, can help speed upward the growth of car learning and accept it to a college level in less fourth dimension.

This move turned out to be the right one because there are then many contributions from the independent developers to TensorFlow that they far surpass Google'south contributions. Wikipedia mentions that "there are 1500 repositories on GitHub which mention TensorFlow, of which 5 are from Google." That being said, i of the discussions at Quora doubtable that the released open-source code is the "cleaned-up" version from the one that Google uses in its services.

How Does TenserFlow Work?

Using the simple normal human language and a heavy simplification, nosotros might see 1 side of TensorFlow equally an advanced autonomous filtering technology. At its eye, the technology is a huge software library of car learning. It uses the database to assist information technology "make decision".

For example, someone uploads a photo to Google Photos. The technology will compare all the details from the picture to its database and decide whether information technology'south a moving-picture show of an animal or human. Then if it's a human, it will try to determine the gender, historic period to all the way to who the person is. The same process is repeated for other objects in the photograph.

Tensor -bb- Cat or Dog

It also uses user's data such as the identity of the person in the moving-picture show and the location where the picture is taken, to enhance its library so that information technology can requite better results in the future – both for the individual who uploaded the photo and for everybody else. Hence the term "learning". Only information technology doesn't cease at just at knowing and learning data from photos. There are and so much that the applied science can do with information from a photograph. For example, it tin can group photos with similar details such equally the aforementioned person, the same location, the same engagement; meet the pattern of faces to determine which family unit and friends the person in the photo belong to, and utilise the information to make videos of family unit vacation or animation from continuous shots.

That barely scratches the surface of how TensorFlow works, but I hope information technology can give y'all a general film of the technology. Likewise, using merely one instance can't do justice to what it's capable of.

And for all the Artificial Intelligence enthusiasts out there, it's worth mentioning that Google already created a figurer chip technology optimized for machine learning and integrating TensorFlow into information technology. It'south called Tensor Processing Unit (TPU) ASIC scrap.

Tensor -bb- tpu-2.0

Those who desire to larn more than about TensorFlow tin visit its tutorial folio.

Applications of TensorFlow

We are at an early stage of machine learning technology, so nobody knows where it will take usa. Just there are a few initial applications might give us peek at the futurity. As it originates from Google, it'south obvious that Google uses the technology for many of its services.

More on Image Assay

We've discussed the case of using the applied science for image analysis in Google Photos. But the image analysis application is also used in Google Maps' Street View characteristic. For example, TensorFlow is used to connect the image with the map coordinates and to automatically mistiness the license plate number of any car that's accidentally included in the image.

Speech Recognition

Google is likewise using TensorFlow for its vocalisation assistant voice communication recognition software. The technology that allows users to speak out instructions is not new, merely including the ever-grown library of TensorFlow into the mix might bring the characteristic upwards a few notches upward. Currently, the speech communication recognition engineering science recognizes over 80 languages and variants.

Dynamic Translation

Another case of the "learning" office of machine learning technology is Google's translation feature. Google allows its users to add new vocabularies and fix the mistakes in Google Translate. The e'er growing information can be used to automatically detect the input language that other users desire to translate. If the motorcar makes mistakes in language detection process, users can right them. And the motorcar will learn from those mistakes to improve its future performance. And the cycle continues.

Alpha Get

I fun example of TensorFlow usage is Alpha Get. It is an application that is programmed to play Go. For those unfamiliar with Become, it'south an abstract board game for 2 players originated in China more than five thousand five hundred years agone, and information technology's the oldest board game that is still continuously played today. While the rules are uncomplicated – to environment more than territory than the opponent, the game is incredibly complex and, according to Wikipedia: "possesses more possibilities than the total number of atoms in the visible universe."

So, information technology's interesting what a learning motorcar technology tin do with the infinite possibilities. In its matches against Lee Sedol – the eighteen-time Go earth champion, Alpha Get won 4 out of 5 games and was given the honorary highest Go grandmaster rank.

Magenta Project

Another interesting application of TensorFlow is the Magenta Project. It's an ambitious project to create machine-generated art. One of the early tangible results of the experiment is the 90 seconds pianoforte melody. In the long term, Google hopes to generate more advanced machine-generated fine art via its Magenta project and build a community of artists around it.

On February 2016, Google also held an art exhibition and sale in San Fransisco showing off 29 computer generated – with a picayune help from human – pieces of art. Six of the largest works were sold for as much every bit $8,000. The computer might still have a very long way to become earlier it tin can imitate a real artist, but the corporeality of coin people are willing to pay for the fine art bear witness united states how far the technology has gone.

Tensor -bb- Google Image Generated Art

Support for iOS

While we take already seen the capabilities of TenserFlow on Android, with its latest version, TensorFlow finally adds supports for iOS devices. Since at that place are tons of great mobile apps available exclusively for iOS, or released first on iOS, it ways that we can wait more than neat mobile apps adopting machine learning in the near future. The same thing can be said for the possibilities of wider adoptions and applications of TensorFlow.

The Future of TensorFlow

What can one possibly do with a machine that is able to learn and make its own decision? As a person who deals with more than than ane languages as part of the daily life, the first thing that pops into my listen is language translation. Not in the give-and-take by word level, just more in the longer text level like documents or even books. Today's translation applied science is limited to the vocabularies. You tin hands find out what is "sleeping" in Chinese and vice versa, but try to throw in ane chapter of Eiji Yoshikawa'southward Musashi in its original Japanese and interpret the chapter into English. You'll see what I'm getting at.

It's also fun to see what the future of Artificial Intelligence can do with music. While it's still very basic, Apple tree's Music Memo app can already requite automatic bass and drum accompanion to your recorded singing. I call back one episode of a SciFi TV show where a character in the evidence created a machine that analyzes all the top songs in the charts and able to write its own hit songs. Will nosotros ever arrive there?

And as the closing thought, I'd like to mention Sunspring. Information technology's a short scientific discipline fiction movie written entirely by an AI screenwriter that called itself Benjamin– which fifty-fifty composed the popular-song musical interlude. The film was put together by director Oscar Sharp for the 48-60 minutes Motion picture Claiming of Sci-Fi London event.