The matter on the sea soon reached Chen Nian's ears, but he was not surprised by such a commotion.

Isn't it the basic operation of all countries to take the warships that were supposed to be decommissioned for the fishery administration and the coast guard?

It's just an international practice, it's not something to be proud of!

What's more, the 051 used in this "transformation" is not old, according to his memory, he will be retired in 06, and before he retires, he will play the residual heat and temporarily save the field, which is nothing.

However, for him, after this incident, the island-building matter was put on the agenda, the obstacles were basically removed, and the remaining work was to build our own Sky Whale in the shortest possible time.

Chen Nian did not go to the scene to command, because compared to all the previous projects, in fact, the overall technical difficulty of the Sky Whale was not as high as imagined.

Its difficulty is mainly reflected in the difficulty of the project.

In other words, the difficulty lies in how to dismantle and distribute a huge project in an orderly manner, how to mobilize the productivity of various departments, and how to carry out process-based and standardized production.

In these aspects, in fact, Chen Nian can't help much.

All he can do is use his own strength to help them break through quickly when the project team encounters a technical level.

So, in the absence of the scene, the difference is not much.

So, during this time, Chen Niandu stayed at Xi'an University of Technology and continued his studies.

He had already completed all the courses of advanced engineering materials, but unlike those written in the fantasy, after learning a complete "secret book", he did not immediately obtain powerful "exercises".

Because the accumulation of this thing is step-by-step, there is no chance to reach the sky in one step.

However, in retrospect, the number of source points provided to him by this whole period of study has also reached more than 15 points, and overall, it is still extremely cost-effective.

In the next step, Chen Nian plans to continue to study material surface/interface science to further complete and enrich his knowledge system of materials science.

The course has been booked, and the tutor is still Mo Zhongmin, with his level, teaching Chen Nian is quite handy

In addition to attending classes, the most Chen Nian is currently doing is to discuss the issue of unmanned aerial systems with Wang Jiancheng.

This was originally a project built by Wang, but after seeing the potential of the project, Chen Nian couldn't help but participate in it.

It's not that I want to pick the fruit of the king's construction, but it's just out of a kind of thing. The curiosity and conquest of researchers.

He wondered if he was capable of making a breakthrough without the help of the system.

This kind of thinking is the same as the elite in the high-ranking workplace, who constantly go to the naked job search.

It's all about testing your true strength and avoiding boiling the frog in warm water and dying in the pot in the end.

So far, he has been pleasantly surprised by the progress.

Although Wang Jiancheng is indeed an out-and-out genius in this regard, his guidance and assistance have also played an irreplaceable role.

After fully understanding the principles of the convolutional neural network algorithm, it took the two less than three weeks to make the first version of the weak AI training model and prepare to start the initial testing.

As an important project partner, Chen Nian will naturally not be absent from this significant test.

After a day of small classes, Chen Nian and Wang Jiancheng made an appointment in the laboratory of the school's computer comprehensive education center, where there is a small computer that can support the computing power required for the convolutional neural network model, which can be used for this training experiment.

——

Of course, being able to approve it was actually the result of Chen Nian's secret arrangement through Chen Guo.

Otherwise, although Chen Nian himself can come and go at will, but as Wang Jiancheng is an ordinary student, no matter how awesome he is, this thing will not be open to him.

Chen Nian sometimes couldn't help but feel a little emotional.

Scientific research is like cultivating immortals, and wealth and wealth are indispensable.

He walked all the way to the direction of the laboratory, and after entering the door, Wang Jiancheng had been waiting there early, and as soon as he saw him walk in, he immediately stood up and greeted warmly:

"Senior! Over here! ”

Chen Nian looked over and saw that there were not only Wang Jiancheng present, but also Chu Jianshu, a professor at Western University of Technology who he had asked for help before, who specialized in teaching flight control.

At that time, Wang Jiancheng planned to start the project, but the resources and information preparation in all aspects were not sufficient, so Chen Nian contacted Chu Jianshu through Chen Guo's relationship and asked him to be an interface.

It is also through his relationship that this laboratory with a small computer will be opened.

The other party is not a member of the Spark team, and he doesn't know Chen Nian's identity, in his eyes, Chen Nian, like Wang Jiancheng, is just an "ordinary" and relatively talented student.

He hurried up to say hello.

"Professor Chu, are you here too?"

Chu Jianshu nodded with a smile and replied:

"Your project is quite interesting, I have also watched it before, and now that it has produced results, I will definitely come and see it. It doesn't affect you, does it? ”

"It doesn't affect it, it's not affected, you're willing to guide, we're too late to be happy."

Chen Nian replied with a smile.

On the other hand, Wang Jiancheng has already set the basic parameters of the model and connected to the flight simulation program.

In the subsequent training, the model will automatically start learning according to the random instructions entered by the flight program, and gradually find the strategy with the greatest benefit with high-frequency trial and error.

"So how did you set your criteria in the end? Or is it still the same as I said before, with the following distance as the standard? ”

Chu Jianshu asked.

"No, it's changed. This criterion is too rigid, if only the distance is used as a criterion, although the success rate is high, but at the same time, once it fails, the consequences are also serious. ”

"We can't add the estimated loss as a criterion, and then the whole model will become too big."

"Even if our small aircraft can run now, we won't be able to run if we really get on drones in the future."

"So, the final strategy is to use the success rate of the return to the team as a criterion of sovereignty, and distance is only used as a support."

Hearing Chen Nian's answer, Chu Jianshu's eyes lit up.

It's a clever strategy.

As an important part of the formation, there is no doubt that UAVs are to form a flight formation with manned aircraft, but due to the difference in performance, the formation of the formation cannot remain stable forever.

——

Not to mention the mixed formation of manned and unmanned, even if it is a purely manned aircraft formation, it is impossible to always maintain a formation in battle.

But on the other hand, if you want to form combat effectiveness, formations and groupings are necessary.

This requires the pilot to return to position quickly after completing the tactical maneuver.

Taking this as a standard to constrain the action of the drone and use it as a criterion for machine learning training, on the one hand, it gives the model greater freedom, and on the other hand, it can also meet the needs of actual use, which can be said to be the best choice in the compromise situation.

These two boys are worthy of geniuses, one law can be used for ten thousand methods, for this model, what they have mastered is definitely not just a single artificial intelligence knowledge.

Object-oriented development is really useful development.

However, how can an ordinary student have such insight?

——

He didn't know that Chen Nian's understanding of Air Force tactics at this time actually had practical experience

"Clever, this should be the best solution for us. Shall we start now? ”

"Let's get started!"

Wang Jiancheng nodded, and tapped the enter key with his finger.

As the fan of the minicomputer turned, the program immediately began to run.

Data is constantly fed into the model, and since there is no visual interface, all the learning process is presented in the form of data.

Chu Jianshu didn't quite understand it, but Wang Jiancheng kept explaining while staring at the flickering lines of code crazily:

"I've completed 1,000 training sessions, the effect is very poor, the success rate is 0%, but it's normal, I estimate that it will take at least a million more training sessions to achieve the initial effect."

"Five thousand times, this is considered a success, a drone has returned, but this is luck, the strategy has not been formed"

Time passed minute by minute, and as the program continued to run, Wang Jiancheng's face became more and more solemn.

There is no other reason, the success rate is too low!

The training effect of this model is far weaker than he expected, and if this efficiency is followed, I am afraid that hundreds of millions of training will not be enough.

What a?

There is an iron law in machine learning:

When the training intensity of the AI is greater than the human training intensity, the model is invalid.

Obviously, this model in his hand is like this.

"Not really. This algorithm doesn't seem to be up to the standard that can be used. ”

"The training efficiency is too low, and the speed of policy generation is slow."

"If we continue to do this, we might as well manually traverse the custom strategy."

Hearing his words, Chen Nian also frowned.

He asked:

"Can you see why?"

"I can't see it. I don't think our algorithm should be a big problem, but cold starts consume too many resources. ”

"The nature of machine learning is also traversal, but it can't be used as a reference for any perceptual experience."

"That is, a priori the human and the machine a priori are not aligned"

Looking at Wang Jiancheng's slightly disappointed eyes, Chen Nian continued to ask:

"What about the confusion matrix we talked about last time? Can it be solved? ”

"It can't be solved. The problem is that there are fewer priors and experiences, and if we want to get a high-level solution efficiently, we must require a very high level of intelligence in the machine. ”

"And at present, the generalization of this task, that is, the difficulty of task generalization, is very large, which further increases the requirements for the machine."

"No, it's not a problem to run like this."

As he spoke, Wang Jiancheng directly knocked the stop button, the model stopped running, and the final result, in more than 60,000 trainings, the number of effective results was only twice.

This amount of data is simply not enough to support the model to continue training.

Wang Jiancheng looked at the results on the screen with disappointment, and at this moment, he felt an indescribable sense of frustration.

In fact, at the very beginning, he was extremely confident in the model.

After getting Chen Nian's click and gradually understanding the basic principles of convolutional neural network algorithms, he almost felt that the whole world was in his hands.

The potential of this set of algorithms is so great that as long as we continue to develop it, it is not a luxury to completely change the world one day.

Moreover, on this track, I don't know how many steps faster than other researchers, as long as I can come up with usable results and cases, I am likely to become one of the pioneers of a new generation of artificial intelligence together with my seniors.

What an honor, needless to say.

With such confidence, he can be said to be advancing all the way, so fast that Chen Nian was amazed.

He originally thought that things would go smoothly like this, but he didn't expect that the first training results of the model would give him a heavy blow.

It's not just that the results didn't live up to expectations, it's that they weren't available!

If it doesn't meet expectations, you can still get closer to the goal by constantly adjusting and optimizing, but if it is not available, it is really useless to try.

Tearing down and starting over may be the only way.

But on their own. Can you still have that kind of heart?

Chu Jianshu listened silently to the conversation between the two on the side, in fact, this result was also within his expectations.

You know, it's artificial intelligence!

Even if it's science fiction, the prospect of artificial intelligence is decades away.

In this era, if you want to achieve primary artificial intelligence under the current computer computing power, you also need to consider the upper limit of the performance of the terminal, how easy is it to say?

Being able to do this is already a success in a sense.

The next road is still to be walked slowly step by step.

——

Of course, if you want to talk about the solution, in fact, he is not without it.

Previously, because of his interest in the project, he also learned some knowledge about machine learning and convolutional neural networks.

According to the discussion between Chen Nian and Wang Jiancheng just now, the biggest shortcoming of the whole model is the lack of a priori, if a large enough sample database can be implanted in the model, so that the machine can deduce and learn according to the samples in the database, then the difficulty of machine training should be reduced geometrically.

Such a strategy is not easy for others, but it is not difficult for him.

Want data, that's a matter of a couple of phone calls.

Based on their own scientific research purposes, wouldn't it be okay for the military to send some training records of pilots? If you really don't have one, let them fly a few times according to the target logic.

But the problem is that such a solution to the problem is tantamount to pulling out the seedlings for these two children.

This time, I can help them solve the problem through privilege, but what about next time?

If this is an urgent scientific research task, then no matter what means are used, Chu Jianshu will not feel excessive.

But now, it's more of an exploration.

Talents need to be cultivated, and in the process of their leveling, you can give them some help and give them directions, but you must not help him brush the boss

Thinking of this, Chu Jianshu asked quietly:

"So what are you going to do now?"

The two were silent for a moment, and then Wang Jiancheng spoke:

"There are two ways to go."

"Or, I can find a way to contact the model airplane agency and ask them to help us brush up on the prior data."

"But the accuracy of the data produced in this way is relatively poor, and I don't know how much time it takes for secondary processing."

"It is very likely that by the time the data is processed, the time to apply for the special class has already been missed. But as long as you can write the model, you don't lose money. ”

"There is another way, which is to tear the whole model upside down and start over."

"I just thought about it carefully, in fact, there are still a lot of efficiency optimization points in the model that have been overlooked, and after these points are solved one by one, there will definitely be benefits."

"The problem is that if you do this, the underlying logic also needs to be modified, and the amount of work is relatively large."

Wang Jiancheng sighed helplessly and looked up at the ceiling.

"At least it's better than no way, if there is a problem, you have to find a way to solve it. Think about it when we get back. ”

Chen Nian nodded slightly and did not speak.

In fact, what Chu Jianshu can do, he can also do.

But his own original intention of joining this project made him unwilling to use this somewhat "opportunistic" method.

It's a bit embarrassing.

However, is there no other way than this method?

He always felt as if he was ignoring something. What is it exactly?

On the other side, Chu Jianshu nodded with satisfaction after hearing Wang Jiancheng's answer.

Fortunately, the first reaction was not to give up, nor did I subconsciously ask for help, but planned a solution.

This shows that, at least in terms of mental nature, they have basically passed the test.

So, he comfortingly patted Chen Nian's shoulder beside him and said:

"It doesn't matter, scientific research is like this, sometimes even hundreds of failures can be exchanged for a success."

"Take your time, you still have time."

"Understood, Professor Chu. Don't worry, this is all a small problem, and tomorrow we will talk to the teachers of the mathematics institute to see if we can do some optimization on the algorithm. ”

"It's good to have this idea! Let's go, let's go have a late-night snack? I invite you! ”

Hearing his words, Chen Nian smiled slightly, stood up with kindness, and did not forget to remind Wang Jiancheng to put the initial settings of the minicomputer back in place.

After cleaning up, the three of them walked out of the school gate together and ate a casual meal, and in the process, both Wang Jiancheng and Chen Nian were a little silent.

Of course, Chen Nian was thinking about the fleeting aura that flashed in his mind just now, as for Wang Jiancheng. Most likely, I am also thinking about the model.

Seeing the performance of the two, Chu Jianshu didn't say anything.

He also thought to himself that maybe the two children were hit and didn't want to speak, and of course they might be planning their next steps, but no matter what, he shouldn't bother them too much.

Things that need to be experienced are always experienced, and since you have made up your mind not to interfere, you have to follow through.

However, when he returned to his residence, he still couldn't help but be a little worried.

Many geniuses, in fact, die in one or several failures.

Especially young geniuses - because they are not mentally mature at this time, the slightest setback may cause a great butterfly effect, and eventually cause irreparable effects.

Will Wang Jiancheng be such a person? Could Chen Nian be such a person?

Chu Jianshu hoped that they weren't, but he wasn't too sure.

——

However, such fears lasted less than two hours.

As soon as he finished washing, his personal mobile phone received a call from Wang Jiancheng.

On the other end of the phone, Wang Jiancheng said excitedly:

"Professor Chu, we still want to borrow a minicomputer tonight! We know how to do it! ”

"What? How? ”

Chu Jianshu asked in surprise.

"That's the case, Senior Chen Nian told me just now, in fact, our thinking at the beginning was wrong."

"We've brought the manipulation of the aircraft into the drone!"

"For us, it's a continuous movement to set the goal, then to turn and move, but that's not the case!"

"From the machine's point of view, the so-called position change is actually a change in data points one by one."

"We don't need to treat it as a continuous function at all, but rather split it into two different models."

"The first model, which deals with location selection, is used to determine location targets."

"The second model deals with policy execution."

"In this way, the complexity of the continuous function will be greatly reduced, and our training difficulty will be reduced by at least 90%!"

Chu Jianshu listened to his words silently, and after being silent for a few seconds, he said sincerely:

"You guys. . . Well done! ”

At this time, an idea popped up in his heart:

Sure enough, in front of a true genius, in fact, there are really not many "setbacks".

(End of chapter)