Japan's artificial intelligence (AI) startup, Preferred Networks, is investing in a field where few people are involved, making plant robots smarter and allowing robots to transfer experiences to each other through special networks to speed up learning. The support of industrial giants.
Preferred Netoworks was founded in just 4 years, and only one product, PaintsChainer, is launched. This software uses AI to automatically color anime. Despite this, Preferred Networks has been heavily sponsored by Toyota Motor, which is expected to generate enough technology to rival Waymo's self-driving, with venture capital firms estimating the company's market value to reach $2 billion.
According to reports, compared to other AI companies, the Preferred Network is different in that there are a number of Japanese industrial giants backing, in addition to the massive data required for deep learning algorithms, but also invested in some rather rare areas.
Preferred Networks founder Ogano Harada and Nishikawa are graduated from the University of Tokyo and are excellent computer science experts. In 2004, the context-aware text classificaTIon won the "Super Creative Home" award from the Ministry of Economy, Trade and Industry.
Nishikawa is interested in computers from elementary school. No matter where he goes, he carries a laptop computer that is the size of a car battery. He claims that it is a notebook, but in fact he uses this computer to write programs at any time. The two founded Preferred Networks in 2014 and decided to focus on industrial robots to make them smarter, and this is an area that Google and Facebook have not yet been involved in.
The idea of ​​Preferred Networks was first appreciated by Fanuc's president, Toshiyuki Inaba. In early 2015, Inaba and Gyanohara and Nishikawa discussed and decided to invest $9 million, they also agreed to let them use Fanuc's most confidential. Some trade secrets are the data collected by thousands of factory robots.
Four months later, Toyota pledged to invest $10 million, and in August 2017, it added $100 million. In December, Hitachi, Mizuho Financial and Mitsui Co successively became Preferred Networks investments. By.
With the support of many industrial giant investors, Preferred Netowrk has access to the world's top factory equipment. Google is able to become an AI power by looking up the engine's data, and Facebook is based on its social platform. In the future, Preferred Networks has the potential to improve the way the factory is manufactured.
Preferred Networks used the Toyota Prius miniature version of the toy car to showcase its technology at CES 2016. The car walked on the road where an obstacle was placed, and almost every time it hit an obstacle, but after 2 hours of training and trial errors, the car had already You can avoid obstacles and whistle, as if you are sitting in professional driving.
In the process, the car does not have any programmers to write instructions for them, but instead establish rules from experience and accelerate the learning speed of each car through a network sharing. A few months later, Preferred Networks made another show that their technology might one day make the factory robot a skilled craftsman.
In the case of Fanuc, engineers write programs that allow robots to grab the right things from a mess of objects. It takes an estimated few days, but Preferred Networks can learn in a short day, when 8 robots work together, short The task can be completed in a matter of hours, and if thousands or millions of robots are linked together, the task will be completed faster.
Fanuc, president of Fanuc, said that it takes 10 years to train a technician, and knowledge can only be taught one by one, but if it is a machine expert, the transfer of knowledge will be far-reaching.
Our factory can produce the substation structures, 35KV, 110KV, 220KV, 500KV various steel components, A shape structures.
Substation Structures,Substation Steel Structures,Structure Of Substation,Substation Steel
Jiangsu Baojuhe Science and Technology Co.,Ltd. , https://www.galvanizedsteelpole.com