Chapter 371: Northwest Dunhuang Journey

"I heard that Wang Xu ran away again? Why didn't you let Xiao Liu clean the table?" Professor Luo Congwen came to Wang Xu's office with a big teacup, but saw Director Jiang wiping the table.

Xiao Liu is Wang Xu's new secretary here. Director Jiang introduced him. He graduated from the National Defense University and is now working in the laboratory. His title is laboratory technician, which is actually a secretary.

This kid has been a soldier and received military training. He was recommended to go to university because of his excellent performance. He has a clean family background and works very quickly. He is deeply trusted by Wang Xu.

Although he has not been here for long, he has already secured the position of the boss in the office. Well, there are only him and Wang Xu in the office. When Wang Xu is not around, he is the boss. There is nothing wrong with that.

"Well, he went to the northwest, claiming to be there to see the scenery, but in fact he just went to have fun. How can you take your wife with you to see the scenery? I asked Xiao Liu to go with him, just to be a light bulb, but he was not dead.

He also brought the drone team with him, as if he wanted to experiment there. Let him do it, it's not like he can't experiment in the capital." Director Jiang felt a little headache, and he often complained about Wang Xu's activities to others.

"Drone team, have they got the algorithm?

I came to him for this matter. He told me that he wanted to use artificial intelligence to train flight control, which is different from traditional algorithms. I just sorted out some clues. Where did he get the algorithm?" Professor Luo looked puzzled. Indeed, Wang Xu recently gave Professor Luo a job to study flight control.

"The one made by the Imperial Capital Aviation Institute should still be the traditional fly-by-wire flight control.

He said that the traditional one also has its advantages, with low computing power requirements, easy cost control, easy modification when encountering problems, and rapid product formation. Both legs must be carried out, and there must be no lameness." Professor Luo glanced at the notebook. Wang Xu had explained this matter, and he had a record.

"Well, it is true. The large model algorithm requires too much computing power. Even Tongfang workstations cannot do it. At least it must be a cabinet machine.

It is estimated that it will not be possible to make a chip to take it to the sky in at least ten years." Professor Luo nodded. The most complicated thing about neurons is the calculation. The more parameters, the more complex the calculation, and the greater the computing power required. The parameters required for a four-axis flight control are not too many.

But if it involves automatic obstacle avoidance, dynamic route planning, and even the operation of drone groups, the computing power requirements will increase exponentially.

Even if the training is carried out on a giant machine, and the control itself is just the result of the calculation. But the huge model network also requires complex calculations to get the results, which is the shortcoming of the artificial intelligence algorithm.

If the traditional flight control algorithm is used, the flight effect is actually not much different, but the cost can be greatly reduced, but for various terrain matching, obstacle recognition, flight route planning and other things, it is a bit powerless.

This is also one of the important reasons why Wang Xu started to lay out the artificial intelligence algorithm at this time. It is hard to say whether you can train this thing, but you must have it in the future.

If the equipment and algorithm data are all in foreign countries, then they will really know your every move, which is much more serious than any bottleneck.

The current layout is mainly the layout of talents. There will be a huge demand for engineers in this field in the future, but few universities can really train them. Moreover, artificial intelligence is not just about neurons. Are there really no other ways to go? There must be schools and professors to try.

The laboratory he set up is based on this idea. Currently, all the students are graduate students. Their names are not under Wang Xu's name, but scattered in various departments of Shuimu, but they are all trained here, and even graduate students from other universities are commissioned to be trained here.

Professor Luo has only two graduate students and one doctoral student. After the filming of the movie next year, Wang Xu will also recruit the first batch of doctoral students, with a plan of two people. These people are all seedlings for future research on artificial intelligence.

Schools need to communicate with each other. A large number of Shuimu graduates will go abroad for further studies. At least in Wang Xu's view, this is not a bad thing. They are indeed leading in many things.

But Wang Xu's lab is not like that. This is the world's top lab, so the admissions review here is very strict. Even Wang Xu is not qualified to intervene, and there are no places for international students.

Unlike scientists who are purely theoretical, Wang Xu has too many engineering practices here, and he can use them directly after taking them back.

Pure artificial intelligence theoretical research is not really worrying. Most of the algorithms in this area are public and not very complicated.

But the big data model is not like that. Many times, it really depends on luck to train it. Many of them are tested bit by bit, and they cannot be handed over to others casually.

Anyway, Wang Xu's laboratory now has very strict security and review, which is not inferior to the confidential department of the top digital research institute, but Wang Xu, the person in charge, is floating outside all day, which makes the superiors very headache.

But the superiors really spoil him, just let him do whatever he wants, and the superiors follow him to clean up his ass, so it gives outsiders a false impression that Wang Xu's laboratory seems very ordinary and there is not much difference from the ordinary Shuimu laboratory.

But in fact, who can come and who can't come has already been arranged. Who can enter which room and who can meet are all rules, but they are not exposed to the public.

Wang Xu certainly knew this, so he just accepted it with peace of mind. He didn't plan to completely hide his identity. Even if he had skills and systems, he still needed a lot of communication and a lot of collaboration to complete most things.

He is not one of those math geniuses who can conquer the world with his clever brain. He really doesn't have this ability.

"Are you here to communicate with Wang Xu today? Or to apply for funding?" After chatting for a few words about Wang Xu's gossip, Director Jiang brought the conversation to the main topic. He was also busy.

"I have them all. When I came here, I heard that he disappeared again. There are some features on the flight control model, so I have to discuss it with him.

I don’t have enough data here and I don’t have enough computing power, so I still have to apply. "Professor Luo immediately explained the purpose of his visit. He was really not someone who just came to chat. He came with a teapot in his arms because he wanted to discuss the issue of model design with Wang Xu.

"It's already January. Although he will definitely be back before the end of the month, he may not be able to spare the time. The funds are easy to talk about, but it's useless to give them to you now. He said that an important reason for going to the northwest this time is to collect data. , He said it’s windy over there.”

"Without data, it's useless to ask for funds, so let's wait until he comes back. I've been busy for a year, and I have two days off. I can't let the boss take a day off, and it's not just the employees who are busy." Professor Luo hehe He smiled, with a proud look on his face.

He can really be proud. The person he is now is completely different from what he was two years ago. An internationally renowned expert on artificial intelligence algorithms, one of the founders of neuron theory, and the first person in China to deepen algorithms in this field.

Unlike the professor in Chou Country in his previous life, who had been working hard for many years and only published papers, Professor Luo has made some small achievements in training. His accuracy in recognizing printed English is already quite high, at least not as good as several other artificial intelligence paths. Difference.

Although its recognition algorithms and models currently have no effective applications due to high computing power requirements, some bright prospects can be seen.

In the field of artificial intelligence at this time, convolutional neural networks are not a mainstream research direction. Therefore, although Professor Luo has made many achievements, they are just that in the eyes of the world.

Of course, this is also related to Wang Xu's conscious control of the speed of publishing papers. He still hopes to hide his capabilities and bide his time for a while. They are now overwhelming China in terms of financial resources, equipment, and even talent reserves.

If the prospects of this road are opened too early, it will actually be very detrimental to China. Professor Wang Xu and Professor Luo will have to suppress the progress. Research will make great strides and papers will be published slowly. This is probably the pattern.

Just like the artificial intelligence learning in the flight control part, the paper may not be published in ten years, let alone open source and shared.

At this moment, Wang Xu was in the northwest, among the desert hills of the Gobi Desert near Dunhuang. Against the strong wind, he took Wang Fei with him to observe several cameras flying in the sky.

Of course, it was brought up by the drone. It can be regarded as a joint debugging. If it can still fly stably in such a windy environment, then it can be finalized. As for obstacle avoidance, it is better to rely on people to avoid it first.

Wang Fei was completely wrapped into a ball by Wang Xu, wearing a down jacket, a down hat, and a gauze scarf. Her slim figure was no longer visible at all, but it still couldn't interrupt Wang Fei's excitement when she saw the desert for the first time.

Of course, this is what typical foreign tourists think, what a spectacular, desolate, vast, all kinds of beautiful adjectives. But in the eyes of the locals, it is only cold, brutal and painful.

The northwest is not a land suitable for living. Thousands of years ago, it was an oasis. Now it is not completely green, but a little further away from the water source is the desolate Gobi.

The real ancient city of Dunhuang has long been in ruins. The current ancient city of Dunhuang is actually a film and television city. It was built a few years ago when I collaborated with Xiaozi to film "Dunhuang". Later, it became a film and television city, specially used to film various scenes in the northwest desert. scenery.

"New Dragon Gate Inn" was filmed here, but Wang Xu didn't need the scenery here. All he needed was the desolate desert. In fact, it was the part outside the film and television city, but some of the supporting facilities in the film and television city itself can still be used. Use it.

There is a national geological park here that will be well-known in later generations, with the famous Yadan landform, but it is not well-known at the moment, and the country has not spared any effort to protect these places.

Wang Xu is now preparing for tomorrow's aerial photography, which should be enough to express the sad atmosphere of the apocalypse.

Wang Xu only needs big scenes here, no detailed performances, just a motorcycle running out of smoke and dust on the ground. Therefore, Wang Xu plans to take some material back this time. Although the season is not right, he might be able to Works wonders.

Wang Xu's original plan was to take pictures after spring when there are strong winds, especially during sandstorms. It felt more right. This time, although it was deserted, it was not cold enough. It was not the cold weather. In fact, the temperature was already below zero.

What I’m talking about here is the cold color. The northwest desert under the sun always has a warm tone, regardless of winter or summer, and regardless of the temperature. This kind of color is not very suitable for the somewhat cold tones of the entire movie "The Matrix", but the golden afterglow at sunset is an excellent atmosphere for showing the dawn.

This is ideal for showing Xi'an's position in people's minds, a ray of sunshine in a cold world, with a touch of dreamy sunshine.

This kind of contrast under the lens can convey an emotion to people unconsciously. This is the so-called lens language.

Of course, this is just the simplest one. There are also the use of lenses, filters, cuts, etc., which are all the language of the lens. This is the role of the director. He uses this special language to lead the audience to use it. A dream journey is completed one shot at a time.

Several drones were taken back. The sun had stopped working. The sky was dark and there was a dead silence all around.

It was very windy today. There were almost no clouds in the sky. It was dry and cold. As the last ray of sunlight faded away in the distance, the vast Milky Way gradually revealed its beautiful shape.

This is the charm that belongs only to the wild. People born in cities almost never have the opportunity to see this stunning starry sky in their lifetime.

Wang Fei had laughed at Wang Xu, saying that he looked up at the stars all day long, but he himself had never seen many stars. Wang Xu angrily dragged Wang Fei to the northwest to drink the cold northwest wind and watch the vast Milky Way together.

Of course, it is not necessary to be so self-abuse to watch the stars. Isn't this catching up?

Moreover, they also have sleeping bags made of new thermal insulation fibers and tents made of new high-strength hydrophobic fibers sponsored by the Department of Chemical Engineering. Once you nestle in the tent, there is really no wind blowing in.

The transparent dome is a new type of transparent plastic cloth, which is a modified plastic. Don't look at it as a small piece, but it was temporarily obtained from the laboratory.

The tensile strength of this thing is dozens of times that of ordinary plastic cloth. It is actually a very ideal space material. It is a bit of a waste for Wang Xu to use it to make a tent.

But who made him Wang Xu? These projects were all created by him. He doesn't need to spend money to find someone to make a film. Not to mention making a tent, he can make underwear.

Of course, this time, Wang Xu brought a lot of new good things, not to mention drones, new cameras, new SLR cameras, new lenses and other professional equipment for filming.

Just look at the pile of equipment on him, what lithium battery-driven LED headlights, micro-mechanical chip-driven digital altimeters and compasses, ultra-high-strength safety ropes, aviation aluminum-lithium alloy hooks, and high-strength stainless steel ultra-thin barbecue grills.

Even the pole in Wang Fei's hand is made of the newly made 1400T carbon fiber. Don't look at it. It weighs less than one or two pounds. Once you insert it into the crack of the rock, you can stand directly.

Anyway, adhering to the spirit of using a cannon to kill a mosquito, Wang Xu brought his hiking kit regardless of the cost

——

(I didn't assemble a new bed at all, I just dismantled the old bed and almost broke my back)