Rewrite the Technological Landscape

Although the itinerary for the second day will not be seen until the next day, many people have guessed correctly that the main content of the second day is really about software, and it starts with Yun Zhilian as the core.

In the morning, the conference focused more on artificial intelligence.

Last year's conference Dafeng Group created a milestone in the field of artificial intelligence, so everyone is full of expectations for the performance of Dafeng Group in the field of artificial intelligence this year.

In addition, an AI chip and corresponding AI applications yesterday have aroused the enthusiasm of the audience, and today's more direct topic of artificial intelligence, everyone is very curious about what Gale Group will bring.

The result is, technology sharing. Dafeng Group has developed rapidly in artificial intelligence in the past two years, and has also launched many applications. At this year's conference, Dafeng Group focused on sharing several important algorithms in its hands, which is very important for people in the artificial intelligence industry. Quite a feast.

In this feast, the algorithm that attracted the most attention, and was rated as the most classic algorithm of this conference a few years later, was the deep convolutional neural network algorithm brought by Shen Lin, who named it Tianmu Network.

The theory of convolutional neural network was born in 1962. It was not until 1998 when the LeNet5 network was proposed that the convolutional network entered the real experimental stage. This stage lasted until the AlexNet network was proposed in 2012 and finally laid the foundation for the convolutional neural network in depth. position in the field of study.

Since then, convolutional neural network has become a very classic deep learning algorithm. With the advent of convolutional neural networks such as Oxford University's VGG network, Microsoft's ResNet network, and Google's GoogLeNet network, convolutional neural networks have become popular in the commercial field. shine.

Now, under the guidance of Meng Qian, Shen Lin has found a breakthrough in the convolutional neural network and successfully developed the Tianmu network.

"Tianmu network is a feed-forward neural network." Shen Lin stood on the stage and explained confidently, "It mainly consists of input layer, convolution layer, activation function, pooling layer, fully connected layer and output layer. composition.

This network has three major characteristics, the first is the local receptive field. For general deep neural networks, each pixel of the image is often connected to each fully connected neuron, and the convolutional neural network can connect each Hidden nodes are only connected to a certain local region of the image, so that the number of parameter training can be reduced.

The second is to share weights. The weights corresponding to neurons in the convolutional layer of the convolutional neural network are the same, which can also reduce the amount of training parameters.

The last is pooling. Since the data that the algorithm needs to process is often very large, especially in the visual field, how to effectively obtain the characteristics of the image is what we are most concerned about. During the research and development process, we are considering whether similar image compression can be used In this way, the image is resized through a downsampling process after convolution. "

Shen Lin spent more than an hour explaining his Tianmu network. Under the discussion of industry insiders, Tianmu network will soon show its special industry value, especially in the field of vision, which is a huge artificial intelligence market in the future.

In one morning, Dafeng Group brought a lot of wealth to the artificial intelligence industry, because there is no way to monopolize theoretical things, and Dafeng Group is willing to share theories to help the industry develop. Of course, this kind of sharing does not completely take The core algorithm is taken out, but it is very good for the market to guide the direction, share ideas, and discuss the future. This is also an industry responsibility of large companies.

In the afternoon, cloud computing, another important section of Yunzhilian, became the protagonist, and the keynote speaker was still Wang Jian.

Two months ago, Salesforce announced its financial report for 2008. The financial report showed that Salesforce’s cloud service revenue exceeded 1 billion yuan, which brought confidence to the entire cloud computing market. Salesforce’s financial report is also known as smashing cloud computing. The final blow to the gate.

Since 2007, companies such as Google, IBM, Amazon, and Microsoft have successively launched their own cloud computing platforms. In 2008, although Microsoft's Windows Azure operating system did not perform as well as Google in later generations, it did become the industry's vane at that time, but In the past two years, the cloud computing platforms of major enterprises have had a problem, that is, there is no market. In the final analysis, the technology is immature and the market trust is low.

In fact, as early as the 1990s, Oracle proposed the technology of cloud computing, but it was not called cloud computing at that time, but the technical principle was the same. As a result, it became a wrong technology at that time because the concept was proposed too early, so in 2007 this When the new wave of cloud computing started, the market was very conservative, and everyone from enterprises to users was groping, and no one dared to bet on cloud computing lightly.

Therefore, Salesforce’s revenue of 1 billion yuan has indeed stimulated the market. It was also in 2009 that cloud computing was recognized by the market in a true sense. The era of computing explosion.

But when Wang Jian introduced Dafeng Group’s cloud service revenue last year at the Fengyun Conference at the Dafeng Global Developers Conference, everyone was stunned. Dafeng Group’s cloud service revenue last year reached 66 billion soft sister coins, far exceeding Salesforce.

From more detailed data, it can be found that 85% of the 66 billion revenue comes from the Chinese market.

When American companies and the American market are still exploring the technology of cloud computing, the development of cloud computing in China is obviously faster.

There are two reasons for this situation. One is the firmness of the Dafeng Group. When the companies in the United States are still spending hundreds of millions of dollars a year to test this technology, the Dafeng Group has already regarded cloud computing as the core technology. Research and development, spent nearly 10 billion soft sister coins a year.

As for why he is firm, and whether this technical direction is right or not, the prophet has already given Meng Qian the answer, so why not be firm?

On the other hand, because China's huge market has greater opportunities for experimentation, there are also 10% of the market available for experimentation, and China's market is larger than the US market.

"Gale Group's cloud computing has already taken the lead." When Wang Jian's speech ended, similar statements were everywhere in the venue.

And when everyone lamented the development speed of Dafeng Group in cloud computing, some leaders from other companies began to introduce their products on the stage one after another. They were Ali, Huawei, Kingsoft, Ying Haiwei, Inspur, China Mobile, Huaxia Unicom and Huaxia Telecom are eight companies.

At the site of the Gale Global Developers Conference, eight companies released their own cloud computing strategic layout.

China Mobile will launch the Big Cloud platform, China Unicom will launch the Wo Cloud platform, China Telecom will launch the Tianyi Cloud platform, Alibaba will launch the Alibaba Cloud platform, Huawei will launch the Huawei Cloud platform, Kingsoft will launch the Kingsoft Cloud platform, and Inspur will launch the Inspur Cloud platform. Ying Haiwei launched the Haiyun platform.

At the same time, the servers of Huawei and Inspur serve a large number of cloud platforms for domestic enterprises.

The participants just lamented the leading position of Gale Group in cloud computing, but now they immediately realized that things are not that simple.

Facing the upcoming cloud computing curtain, Chinese companies and American companies seized the opportunity at the same time, forming a cloud computing Huami confrontation trend.

But judging from the development of the leading companies on both sides, the Gale Group is clearly ahead.

Thanks to AsKing for the reward!