Chapter 909 Deep Learning Framework

At 1:00 p.m., the host made a foreshadowing for the afternoon topic.

"For a long time, the operating system has always been the lifeblood of the technology industry. Microsoft has used its own growth to prove to the world that whoever has the right to speak in the operating system will control the ecology. It is also because of the operating system that Microsoft It has been in an industrial monopoly position for a long time.

It was not until the emergence of smartphones that it impacted the operating system industry for the first time. After years of ups and downs, Hongmeng, Vientiane, and IOS have become the three major operating systems in the smartphone field, and have influenced the ecology of the entire smartphone field.

And now, another impact in the operating system field has arrived, that is the artificial intelligence operating system. Today we will focus on the deep learning framework.

The deep learning framework serves as a link between the past and the future, connecting chips and large-scale computer systems to various business models and industry applications. Therefore, many people say that it is the operating system of the intelligent era.

It can be said that the explosion of artificial intelligence is due to the explosion of deep learning, and the explosion of deep learning is inseparable from the right time, place and people. The arrival of the era of big data and the evolution of chips all contribute to the development of deep learning in this era. Bright laid the groundwork.

In addition, the most fundamental thing for the explosion of deep learning is the development of deep learning frameworks.

At present, there are five major deep learning frameworks in the world, namely Deep Wind of Dafeng Group, Caffe born in Berkeley, Tianwen of Pure Land Lab, TensorFlow of Google, and Theano born in Montreal Polytechnic.

However, in the past few years, with the in-depth development of artificial intelligence in various fields, the deep learning framework has opened a round of comprehensive and detailed explosions, such as Tsinghua University's Jitu, Megvii's Tianyuan, Huawei's MindSpore, Baidu Deep learning frameworks such as PaddlePaddle have begun to show their talents and develop rapidly.

Today, we come to experience a feast of deep learning frameworks. We invited more than 20 core developers of deep learning frameworks to come to the scene to explain to us the deep learning framework of the topic and their views on the entire industry. "

After the host made a foreshadowing, more than a dozen companies including Dafeng Group, Amazon, Google, Huawei, and Baidu, as well as core developers of deep learning frameworks such as Caffe and Theano, came to the stage one after another.

Well-known framework developers basically focus on the advantages of their own deep learning frameworks from the three perspectives of core framework, tool components and service platform when introducing them.

And some relatively new deep learning frameworks are basically Huaxia frameworks. These developers will give some explanations for specific fields. In addition, Huaxia framework developers will talk about standardization, automation, and modularization. Promote the arrival of the era of intelligent industrial mass production.

"I feel that Huaxia companies are going to do something about the deep learning architecture?" Listening to the introduction of one company on the stage, the audience began to discuss again, "I always feel that these companies in Huaxia are planning something."

"The Dafeng Group and the Pure Land Laboratory seized the opportunity and occupied an absolute high ground for China in the global development of the deep learning architecture. Now a large number of Chinese companies have come up with mature deep learning architectures, and they are all rushing to promote artificial intelligence. Intelligent standardization, automation and modularization, if you really want to do something together, it is really possible."

"You see, Ali's XDL architecture is used for advertising, Xiaomi's MACE is used for mobile devices

Among Huaxia enterprises, Dafeng Group has fully rolled out deep learning applications, and there are new entrants working in special fields. If Huaxia enterprises really cooperate, it means top-down and bottom-up. Enterprises are responsible for spreading the plate, and some enterprises are responsible for refining the field, and then everyone cooperates and exchanges technical data.

Does Huaxia Enterprise mean to swallow the entire industry line? "

"There is a huge industrial chain behind the deep learning framework. Once it monopolizes the deep learning framework, it will eat up most of the profits of the entire industrial chain, just like Microsoft back then."

"Yes, in theory, once the deep learning framework monopolizes, it can determine the fate of hardware manufacturers.

Google’s ambition is not here before. I remember that Google itself said in an interview that once TenslorFlow monopolizes, as long as they give up GPU, GPU manufacturers such as Yinhe Technology will die directly. "

"Don't talk about giving up the GPU, even if you don't give up the GPU, once TenslorFlow monopolizes, if Galaxy Technology continues to use GPU to develop artificial intelligence, it will be working for Google. Google doesn't need to directly target Galaxy Technology. Galaxy Technology itself will You have to be caught in a dilemma, if you continue to develop, you will always be subject to Google, if you don’t continue to develop, you may have to start all over again, a huge loss.”

"The deep learning framework affects more than just the hardware. Google also said at the time that the reason why it is so obsessed with the deep learning framework is that it can be bound to Google's cloud services later, driving Google's advantages in the cloud."

"I think you are being polite. Think about what Microsoft's monopoly was like at the beginning. Once the deep learning framework is monopolized, the independent development of artificial intelligence by individuals, companies, and even countries may be controlled."

"Google's boasting before now seems unlikely to be realized, but on the other hand, what if Huaxia realizes Google's boasting at the time?

In the future, the development of artificial intelligence in the world will depend on Huaxia's face?

There is no need for Galaxy Technology to be in a dilemma, but should Nvidia be in a dilemma? "

"But do Huaxia companies have this ability? After all, there is only one big wind group on top of it now. It's not so easy for such a big ecology in the world?"

"Friend, are you here this morning?"

"Yes, how right! The prosperity of Huaxia in the application layer and the general AI technology layer will definitely stimulate the ecological explosion of Huaxia's deep learning architecture!"

"It can be said that there is still a long way to go for the development of deep learning architecture. This period of time can range from three to five years to decades. Now it seems that the fastest running is not only the technology, but also the application. Is there a market for Huaxia?

After a few more years of development, Huaxia enterprises will work together to achieve twice the result with half the effort! "

"The deep learning architecture is the key underlying technology for the development of artificial intelligence. If you grasp the underlying technology, you can grasp the industry!"

"Huaxia Enterprises is worthy of suffering, and knows that only by grasping the underlying technology can it be a real artificial intelligence power.

Maybe a few years later, the development of global artificial intelligence in the entire algorithm layer, service layer, and application layer will really be inseparable from China's underlying framework. "

"However, if Huaxia monopolizes in the future, will Huaxia embarrass others?"

"Probably not. Isn't Huaxia always advocating win-win cooperation? From this perspective, it seems that Huaxia is ahead and it's not a bad thing."

"Although Huaxia Enterprises will not embarrass others, Huaxia Enterprises will fight back."

"you mean?"

"Companies in other countries should not have to worry about anything, they are American companies"

As everyone guessed, American companies are in a hurry now.

The conference was still going on, and Microsoft's Nadella contacted Intel's Krzynick, "Have you seen the news of the World Artificial Intelligence Conference?"

"I see."

"The outbreak of smartphones has made Hongmeng and Vientiane the absolute leaders in mobile operating systems. If the deep learning architecture is taken ahead by Huaxia, the era of smart operating systems will have to follow Huaxia!

In today's conference, the ambitions of Huaxia enterprises have been shown! They are growing so fast! "

"So what are your plans now?"

"I contacted many companies, and we talked with many companies, so you should come here too."

"I may not be in the mood to think about Microsoft right now."

"What's wrong?"

Krzanik said calmly, "After the deep learning framework, the next topic of the World Artificial Intelligence Conference is the chip!"

I'm panicking too!