Suddenly learning at scale across domains.

When I don’t learn, I feel terrible.

But once you really learn it.

It will only feel more terrifying.

And what Li Zheng has to deal with is not the exam, but the reality.

His first step is to achieve a bachelor's degree in computer science.

As an experienced programmer, this is not difficult.

The second step is to reach the level of a master's degree in machine learning.

this……

It's actually not that difficult.

However, master's degrees are also divided into master's degrees and master's degrees.

As a master of fishing, if you can pass the exam and graduate successfully, you will almost be able to find a job.

As for a master's degree, you must master a large amount of core knowledge in the major during the study period.

Including but not limited to probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory...

The requirements for mathematics have exceeded the sum of all the competitions Li Zheng has participated in before, and are approximately equivalent to the amount of study required for 80 national scholarships in biology.

Under this level of mathematical pressure, Li Zhengyi's 100 hours of study time was actually not enough, and Zhuangzi should have targeted him even more miserably.

What I never expected was that in this desperate study intensity, a little bit of student movement that was once very unintentional played a role.

[Learning Talent: Mathematical Brain]

[In thinking, the problem can be transformed into a mathematical problem as much as possible, and possible calculations can be completed with terrifying neural efficiency. 】

Li Zheng could only feel this talent through reconstructing mathematical thinking under the guidance of Jie Qifen before when he was doing mental arithmetic and quick calculations. The initial intuitive feeling was just to enhance his calculation ability.

But at this time, with the deepening of mathematical research, I sneaked into various groups and watched some exchanges between some masters and Ph.D.s. I just discovered the cholesky decomposition, Laplacian matrix and normalized Laplacian matrix that have troubled everyone for a long time.

The difficulty, in my case, was just a matter of smooth sailing and it was easily solved.

On the other hand, as my mathematical foundation becomes stronger and stronger, when I look back at the existing algorithms provided by Shiyi, it will naturally become more and more smooth, and occasionally some optimization ideas will burst out.

Unknowingly, what he said when we first met to resolve their disputes did not fade over time, but became increasingly clear -

"Learn mathematics well and learn as much as you can. It is not only a competition, but also important in the future."

I'm studying now, thank you, teacher.

However, even if you get a master's degree from a big shot, it's still only the second step.

The third step is practical application, where you can use this knowledge to participate in work and actually improve the fault diagnosis algorithm.

This requires a lot of project experience accumulation, and it really takes 996 and half a year to get started.

But for some people, when you are still outside the door and crawling in, she will already get in like a monster without knowing when.

[There is a monster. Every time you step onto a floor, you find that she is still above you.]

【——Lin Yujingyao】

After learning less than 1/20 of Li Zheng's content, Lin Yujing reached the master's level before Li Zheng. At 5:50 pm on the third day of the internship, Lin Yujing completed [Engine Fault Optimization Based on xgboost and Autoregressive Models]

], and [Liquid Rocket Engine Fault Diagnosis Based on Comprehensive Fuzzy Clustering Algorithm] were sent to Li Zheng.

"I...I haven't learned it yet..." Li Zheng looked at the pieced together algorithm documents, his hand holding the mouse trembling a little, "Do you know what these things are?"

"I don't know completely, but you don't need to know the details." Lin Yujing kindly helped him pull the document down. "Here are some related papers. I only adopted them based on the conclusions. Thank you for your understanding."

"It takes a week to figure out any of them, okay?"

"Ah, why aren't you excited?" Lin Yujing joked, "Didn't I finish studying three books in one day? I can't learn from you, right? You should learn from me~~"

"... Damn it, let's learn it then." Li Zheng gritted his teeth and said, "Don't lie to me. Are you sure these two directions are feasible?"

"At least it's better than the current ones." Lin Yujing pointed to a row of papers on the screen and said, "The premise is that the data and results provided by these papers are not filled with water."

"Okay." Li Zheng rolled up his sleeves and said, "Try to figure it out today, and then ask the 11th Institute for a test server."

"In addition, it is feasible for you to do PSO optimization within the existing system." Lin Yujing moved the mouse on another document, "These are the papers and technical documents you need, I didn't look at them carefully."

"Enough, enough..." Li Zheng said hurriedly, "Let's take these three paths first..."

"That's it? I'm also planning to work on the negative selection method. Also, have you learned the morphological classification?" Lin Yujing narrowed her eyes, "I can't feed you enough."

"I'm so full that I'm almost exhausted." Li Zheng waved his hand quickly, "You go and play minesweeper for a while... By the way, my mother recently helped with scientific and technological construction in Africa, and my dad went to participate in something... International Neuroscience

What's wrong with surgery? There's been no one at home for three or four days. Wangcai may be starving to death or drowning in his own shit. I have time to take care of him."

"!!!" Lin Yujing collapsed instantly, "You... how could you bully her like this..."

"I also want to go back and take care of it, but only here has an intranet and can connect to the test server." Li Zheng said, taking out a bunch of keys from his bag and slamming them on the table, "Here are the keys for you, help me see if she is still there."

Are there any vital signs?"

"I... I'll go to your house by myself..." Lin Yujing looked at the straight men's key chain with only pure iron rings. After struggling for a moment, she suddenly took it away, "It's all for the cats, don't think too much.

!”

"What time has it been? How can I think of anything else?" Li Zheng returned to the screen. He had just typed two lines and then turned around, "By the way, don't go through my drawer."

Lin Yujing was overjoyed when she heard this and quickly put on her schoolbag: "I won't dig it if I don't do it~~"

"No... your expression has completely revealed..." Li Zheng stood up and tried to grab the key, "Give it to me, I will take off the small key..."

Lin Yujing immediately protected the key and ran away with a kick.

"I will take good care of the cat."

"Fuck..." Li Zheng sat back at his workstation angrily and muttered through gritted teeth, "Forget it...you will know sooner or later anyway..."

…

At the same time, regular meetings in one hospital and one office building were held as scheduled every half week.

The meeting was chaired by Zhao Zhenhua, chaired by Chen Hongbing, and attended by Shen Tinglan and other team leaders.

As usual, each group reports on the progress of zeroing.

As usual, the person with the biggest bags under his eyes, Zhou Chenglong from the technical team, was the first to report.

Although he was sleepy and exhausted, he still brushed off the broken hair on his scalp and spoke loudly with enough energy.

"The data analysis work of the technical team is nearing completion, and 87% of the planned progress has been completed, which is about one and a half days ahead of the planned time."

"So far, we have jointly proposed two high-probability failure models and four medium-probability failure models with the Sixth Academy."

"Among these six fault models, five are pure engine faults. The Sixth Institute has tested four of them, and one of them has basically completed the reproduction of Huang Er's fault."

"The last fault model is caused by the overall design of the rocket. We need to start the test at the western suburbs test site and apply for test start preparation here."

Zhao Zhenhua suddenly asked: "What does it mean to basically complete the fault reproduction? Explain 'basically'."

Zhou Chenglong gestured: "In the repeated fault, the engine parameters basically matched the situation before Huang Er stopped, but they were not completely consistent. After all, the actual lift-off situation was more complicated, and the data of each lift-off

It will all be different.”

"Then let me ask another question." Zhao Zhenhua then asked, "What is the probability that this fault model really matches the previous accident?"

Zhou Chenglong muttered: "More than 60%."

Zhao Zhenhua's expression dropped visibly. He wanted to say something, but he also knew that it was not easy for everyone and did not want to hurt anyone.

"Secretary Zhao..." Chen Hongbing explained from the side, "Because Huang Er's rocket body has not been salvaged, we can only use various models to recreate the accident. Just like the black box in the airplane accident, we can't get the rocket.

With a black box, it is never possible to determine the cause of the accident 100%. 60% is indeed low, but the limit is only about 80%."

"The question is, how did you get this 60%?" Zhao Zhenhua clapped his hands and said, "When we face the zeroing review team, we can't say it's Zhou Chenglong's personal judgment, right?"

When Chen Hongbing heard this, he quickly winked at Shen Tinglan and asked her to speak.

Shen Tinglan spoke at this point.

"There is a set of passing standards for the evaluation of fault recurrence."

"To put it simply, it is to mathematically compare the data from the fault reproduction with the data from the actual accident, and finally obtain a reproduction accuracy rate."

"According to the data standard for fault recurrence, it can only reach 60% for the time being."

"But for me personally, I am basically 99% certain about this accident model."

"I understand that." Zhao Zhenhua nodded and said, "I have seen this set of accident models. It is indeed very intuitive and consistent with my many years of aerospace experience. But the problem is that only 60% of it meets the review standards. Even if it is really

It is the cause of the accident, but if it fails the review, the work cannot continue.”

"It doesn't necessarily mean that you can't pass..."

Just as Shen Tinglan was about to say something, Chen Hongbing suddenly raised his hand, interrupting her, and turned to look at He Zhongrui.

"Everyone has been working very hard recently. Let's talk privately about the technical team later. Let the test team report first. After the report, one group will go to another group and seize the time to rest."

Shen Tinglan said nothing more, and Zhao Zhenhua also understood.
To be continued...