Home > News content

Behind the Google open source systems have what doorway TensorFlow

via:博客园     time:2016/7/21 10:00:22     readed:1347

As Google's most important artificial intelligence system, powerful TensorFlow function. Some tools included can automatically identify sounds and images, while other tools you can understand the meaning of words based on context. Also, this is an open source system that allows anyone to download the source code. With all the open source tools, TensorFlow allows developers to create highly intelligent products.

Depth learning algorithm

AI is not a novel concept. In 1956 the first conference, the father of information theory of Claude & middot; Arrowood & middot; Shannon (Claude Elwood Shannon) put forward the concepts of artificial intelligence on the machine, and is expected to solve the problem within 20 years .

Obviously, this did not happen. Facts have proved that the human mind is too complex and can not use simple logic statements coded. Many computer scientists have tried to make the computer a professional thinking, but on the information to understand there will always be a huge error. Simply put, based on the rules of logic machine system does not have the human intelligence.

In contrast, similar TensorFlow like artificial intelligence system itself is capable of learning. So Google also continually improve their algorithm open source TensorFlow way that allows a limited knowledge of artificial intelligence developers can develop useful applications.

In fact, it TensorFlow not perfect, but it can help users to filter the information, saving a lot of time. Feedback mechanism which includes also can let the system itself constantly learning from our mistakes to improve. Google has yet to do anything to create artificial intelligence system, but to strengthen the effective coordination of human and machine.

Open source TensorFlow

Decision was made open source Google chief scientist Jeff & middot; Dean (Jeff Dean), he believes the progress of work under the general procedure for innovation too slow. Google is often a developer to write code, or even to wait a few months before discussion at a meeting, in addition to another developer will then spend several months repeat the process.

Dean believes that open source TensorFlow can significantly accelerate the process. Open-source, Google developers can collaborate in real time with the scientific community. Personnel other than Google can be involved in the preparation TensorFlow source code. The machine learning techniques can be shared widely to attract more technical personnel to improve TensorFlow system.

TensorFlow project leader Rajat Monga pointed out, & ldquo; By TensorFlow open source, we are able to start-ups and universities, and many developers to cooperate, exposure to new ideas, promote technological development. Open source makes code development faster, TensorFlow more versatile, flexible and convenient. & Rdquo;

Build the value of ecosystems

From the traditional concept, the Google open source machine learning tools seem very strange. We can not imagine Coca-Cola will announce its formula to the public. Including Apple and many high-tech companies are tight-lipped about the new product. Even Google, many shrouded in secrecy, such as its search algorithm.

Yet the world is changing. Surest way to success is to have a unique value chain optimization. By sharpening internal processes and expand its scale and continuously improve their dominance among the customers and suppliers to create greater efficiency. This is the companies build their competitive advantage in major ways.

Today, the most successful product gradually transformed into the value of the ecosystem. Although Google is hiring talented people, but it is the only way to improve technology and larger scale the entire scientific community to cooperate. Through open source, you can use the majority of social art using Google technology to develop products.

That is why many high-tech companies will open source its key technologies. In addition to its open-source Google TensorFlow addition, Facebook also announced its open-source library of artificial intelligence tools, electric vehicles, Tesla also open its patents, but also open source recent IBM quantum computing platform.

Today, we live in the era of the Internet economy. Cost reduction and asset optimization is no longer an absolute competitive advantage, and value of ecosystem is to create a competitive advantage in the true sense. Corresponding capacity advantage is no longer top of the pyramid, but in the network center.

China IT News APP

Download China IT News APP

Please rate this news

The average score will be displayed after you score.

Post comment

Do not see clearly? Click for a new code.

User comments