Luo Sheng returned to the company headquarters and started a new plan in the office where he was located in Blue Star Technology. He is probably the only one who can complete this plan, and he just wants to do it himself.

Developing AI requires supercomputers. Luo Sheng does not have private supercomputers now, and the two companies do not have their own independent supercomputers, unlike ibm.

However, Luo Sheng now has data centers and cloud computing, and the underlying layer of cloud computing is also a supercomputer. Through the data center and operating system, super-large-scale physical servers in different regions are aggregated into a supercomputer for unified dispatch and use.

There is no such thing as the two sides, because each has its own strengths and different focus. With the development of technology, cloud computing and supercomputing will become a powerful technology engine for the big data era.

...

At this moment, Luo Sheng created a new folder on one of his working computers, and the index directory is called "nov". This is the code name, or name of the artificial intelligence to be developed.

According to the pronunciation of nov, it is translated directly with its homophonic meaning, and the Chinese name is "Nafu".

After confirming the name, Luo Sheng began to program and develop Nafu's agent.

If you want to develop artificial intelligence, you must first understand three concepts: artificial intelligence (artificial intelligence, abbreviated as ai), machine learning (mae learning, abbreviated as ml), and deep learning (deep learning, abbreviated as dl).

These three are closely related. Simply put, artificial intelligence gives human intelligence to a device or machine. Machine learning is a method to realize artificial intelligence, and deep learning is a technology to realize machine learning.

Artificial intelligence includes machine learning, and machine learning also includes deep learning.

Last year, in 2006, Geoffrey Hilton, a professor at the University of Toronto, Canada and a master of machine learning, and his student Lasla Hudinov published an article in Science, which kicked off the wave of deep learning in academia and industry.

...

In the following days, Luo Sheng established a terminal-to-terminal "ecosystem" for Developer Nafu and personally wrote programs for it.

Luo Sheng has not taken much care of both companies recently and has been a hand-off manager for a while. On the one hand, the company is not busy now, and the arrangements are clearly clear. Secondly, there is Yao Jianhong on the Coast of Blue and Zhang Bowen on Blue Star Technology.

This "ecosystem" is currently being arranged in the underground warehouse of his private villa. When his super luxury house is completed, he has moved into the past as a whole. Luo Sheng has now spent billions to build a luxury house with a strong sense of technology, located on the shore of Yangcheng Lake.

Developing artificial intelligence is not achieved overnight. It is a continuous upgrade and optimization path, from simple programs to weak artificial intelligence, strong artificial intelligence, and even upgrade to super artificial intelligence.

The upgrades and advances include the advancement of software and hardware. Instead of typing a few lines of code on the keyboard, a smart AI with self-emotional logic analysis was born.

The "ecosystem" created by Luo Sheng includes the design of a simulator, a set of programming languages, an integrated programming environment, an algorithm library, a set of tools for deep learning using firmware, a set of teaching logic courses, and cloud computing support.

That being said, this "ecosystem" is not enough.

To achieve truly strong artificial intelligence, more advanced and more hardware support is needed, so that we can create perception, action and cognitive abilities that can simulate the human brain. High-level abstraction processing of data can be performed using multiple processing layers containing complex structures or multiple nonlinear transformations.

Obviously, deep learning algorithms are extremely critical.

Another key link is data. Only companies that master massive amounts of big data have the advantages or foundation for developing artificial intelligence.

There are actually not many companies or institutions in the world that have this advantage. There are Google, Yahoo, ibm, etc. abroad, and there is no doubt that it is "l-bat" in China, especially "l", that is, Blue Star Technology Group.

The amount of data it has mastered can be said to be one of the best in the world, including bluespace, cloud song search, reddit, bluestarvideo, etc. Bluestar Technology's business sectors are filled with vast sea of ​​data.

However, the "nov-Nafu" that Luo Sheng wants to develop does not belong to Blue Star Technology or the Costa del Blue, and is completely independent and belongs to Luo Sheng's exclusive technology, but it is certain to provide AI technical support to these two companies in the future.

As the founder of Blue Star Technology Group, Luo Sheng’s database is undoubtedly completely open to him. The upcoming “nov-Nafu” will enter the database of Blue Star Technology Group and the Costa del Blue.

...

Time passed day by day, and Nafu's prototype also completed the pioneering stage after Luo Sheng stayed up late and fought continuously. Although it was far from the true artificial intelligence, this was a breakthrough between 0 and 1.

No matter what, we have to accept that there are really geniuses in this world. Compared with others, people will only make people angry.

What's even more annoying is that genius people work harder.

Early this morning, Luo Sheng came to the company, but he did not handle the company's affairs. He stayed in the office and was trying to make trouble for Nafu. The original code had been written.

Luo Sheng opened his dedicated work computer to link to the cloud, located in the Sucheng Data Center, and his private cloud service.

The index quickly reached the "nov" folder, and after confirming that the operation was correct, it started without hesitation and woke up "Nav".

Seeing the percentage progress bar continue to advance, Luo Sheng waited patiently.

However, he also knew that even if Nafu was awakened now, she would be a "artificial intelligence retarded". The road would be taken step by step. Upgrading from "artificial intelligence" to "artificial intelligence" and then upgrading to "strong artificial intelligence" or even to the ultimate form of "super artificial intelligence", is not achieved overnight.

Completing the first step is often the most difficult.

【100%, startup is completed!】

"Nice!!"

Luo Sheng, who was staring at the screen, was immediately excited when he saw this scene. He immediately connected to the external network without saying a word. He had already made a "plug-in".

At this moment, Luo Sheng opened Blizzard Battle Network and quickly registered a Battle Network number with the ID nickname "nov".

Then, he spent a few minutes to arrange what Nov was going to do today, which was to let "she" attack the qualifying match in Blizzard Battle Network's "Starcraft: The Battle of the Mother" ladder system.

Compete with those top players.

This is the first learning course that Luo Sheng determined for Nafu, allowing "she" to promote deep learning for research on real-time strategy games (rts).

StarCraft is essentially a fast-paced war game. Players play the role of commander who controls the overall situation, so any operation, decision, and consciousness of the player will directly or indirectly determine the outcome of a confrontation.

Players need to always consider the balance of economic and military development, and make a lot of decisions in a very short time, such as violent soldiers, mining, climbing technology, offense, defense, sneak attacks and anti-sneak attacks, harassment and anti-harassment, reconnaissance and anti-reconnaissance, etc.

Such a real-time strategy game is undoubtedly an excellent platform for nov to advance, but Luo Sheng gave Nafu certain restrictions before "her" entered the celestial system to compete with those top human players.

First of all, there is the fog of war, exactly the same as human players, or they will open the map and hang it up.

The second is to limit the nov to APM and EPM. If there is no restriction, human players will lose 100%, and it is impossible to beat it. This is obviously not a fair confrontation, nor is it Luo Sheng’s original intention, and it will not be of much help to nov deep learning training.

The so-called APM is the number of operations per minute of players. For example, the top star players in South Korea can soar to a terrifying 500 on average, while APM is the number of effective operations per minute of players. APM is counted as long as the player clicks on the mouse and keyboard, and it will be counted as soon as the player clicks on the mouse and keyboard. Therefore, there are a lot of waste operations in the apm, and each operation in the epm is an effective operation. For example, if the behavior is repeated, it will only count once.

Luo Sheng set a 550 APM upper limit for Nafu, which can achieve the same value as the world's top interstellar professional players, while the online epm is set at a 188 value.

At the operation level, Nefu is restricted and she is in the same position as the top professional players of humans. If "she" wants to defeat the top players of humans, she needs to be exerted at the "intelligence" level. To be down-to-earth, consciousness is very important.

After Luo Sheng got everything done, he shook his spirit and decisively put Nafu on the battle network, allowing "her" to enter the celestial system to match the players to fight.

In Luo Sheng's eyes, Blizzard's Battle Network system is like a slag. It has 100% absolute confidence to ensure that the "glass" system cannot detect any abnormalities in Nafu, so she disguises herself as a human player to fight against real human players.

At the same time, Nafu also connected to Blue Star Technology's database and continuously contacted and analyzed massive amounts of information through deep learning algorithms, which would not prevent her from competing with human interstellar players on the ladder at the same time.

This is how humans' unparalleled advantages are reflected in comparison with AI.

Next, Luo Sheng did not interfere, and sat motionlessly, staring at the computer screen.

Now that his computer has been taken over by Nefu, he entered the ladder system to start matching players, the battle mode is the 1v1 confrontation mode with the highest competitive level content.

Luo Sheng didn't say a word at this moment. He quietly watched Nafu's first perspective. It took less than ten seconds to search for the opponent and quickly found the player.

The first game of confrontation officially began.
Chapter completed!