I control the AI ​​empire thanks to the geniuses in my group.

The curling test is done.

Su Zhou turned to the first page and glanced at it.

At the same time, a very faint "Holy shit!" came from behind him.

Similar sounds gradually rose from various corners of the classroom.

Teacher Zhou stood on the podium, outwardly calm, but inwardly closely observing the reactions of each student.

His gaze inadvertently swept over Su Zhou.

The student who had always intrigued him was now sitting quietly in his seat looking at the exam paper.

Then Su Zhou picked up his pen, without turning to the later, more challenging questions to check their difficulty, and without even unfolding his scratch paper, and went straight to the first multiple-choice question.

Teacher Zhou raised her eyebrows slightly.

He noticed that Su Zhou's way of solving problems was somewhat unusual.

Most students' process for answering multiple-choice questions is: read the question → think → do the calculations on scratch paper → eliminate options → fill in the answer.

But Su Zhou's pen almost never left the answer sheet. After reading the question, he would pause for no more than a few seconds and then fill in the answer directly.

Teacher Zhou frowned and stepped down from the podium.

He slowly strolled to Su Zhou's side, pretending to be inspecting her, while glancing at Su Zhou's answer sheet out of the corner of his eye.

The answer to the first question is B.

correct.

The answer to the second question is D.

Teacher Zhou did a quick mental calculation and confirmed that it was correct.

For the third question, the answer is A.

correct.

Question 4...

All six multiple-choice questions were answered correctly.

Teacher Zhou paused for a moment.

These multiple-choice questions weren't just randomly selected.

He deliberately designed several distractors when setting the questions. Without a solid foundation and a clear problem-solving approach, it is easy to fall into the trap.

Especially the fourth question about the range of functions, he originally estimated that the class's correct rate would not exceed 40%.

Su Zhou filled in the correct answer in less than ten seconds.

No draft paper was used.

Teacher Zhou had a question in her mind. She subconsciously turned around and glanced at the other students' papers. The one on the left got three out of the first six multiple-choice questions wrong.

I got stuck on the third question in the next one. I wrote a whole lot on the scratch paper and then crossed it all out.

Teacher Zhou was a little annoyed.

No matter how well Su Zhou answered, it only shows that Su Zhou is talented, but it does not prove his teaching results.

The teaching quality of a class cannot be guaranteed by a single genius.

After looking around, he realized that the results of his hard work teaching for more than half a month seemed to be somewhat... not very lush.

As he was thinking about this, he suddenly felt someone watching him.

Looking down, he saw that Su Zhou had stopped writing at some point, turned around, and was quietly looking at him.

Then he raised his hand.

Teacher Zhou walked over, bent down, and asked in a low voice, "What's wrong?"

"Teacher Zhou, may I hand in my paper early?"

Teacher Zhou was stunned.

He glanced at the clock on the wall subconsciously. Only thirty minutes had passed since the exam started, and the two-hour exam was still less than halfway over.

Then he looked down at the test paper in Su Zhou's hand.

He filled all the multiple-choice and fill-in-the-blank questions with his answer sheets, and densely filled the answer areas for the four major questions with his solutions. Even for the last and most challenging question, a comprehensive application problem that he adapted from a provincial competition question, he wrote out the complete solution steps.

Teacher Zhou nodded helplessly.

Su Zhou stood up, tidied up the test paper and answer sheet, and placed them on the podium. Then she returned to her seat, picked up her schoolbag, bent down, pulled a history textbook from her desk, and put it in her schoolbag.

All eyes in the classroom had unconsciously turned to them.

The student sitting in the third row glanced at Zhong subconsciously, then looked at his own test paper where he had only gotten to the eighth multiple-choice question, and his face instantly fell.

The student sitting by the window trembled, drawing a long line on the test paper with his pen.

Damn it, you people in the math Olympiad class, are you trying to kill us?

But if we were to name the person who was the most upset in the entire classroom, without a doubt, it would be Su Zhou's deskmate, Zhao Zixuan.

He had just finished the multiple-choice questions when, out of the corner of his eye, he saw Su Zhou had already stood up and handed in his paper, and his blood pressure instantly soared to 180...

After the math exam, the atmosphere in the classroom was like that of someone who had just experienced a massacre.

In the corridor, students gathered in small groups, each one louder than the last.

"What the heck is that third big question? I didn't even understand the question!"

"Was that last, most challenging problem even humanly possible? I filled an entire page of scratch paper just writing out the equations!"

"Does the teacher who set this question have a grudge against us? Why didn't they just make us do the college entrance exam questions instead of this difficult ones?"

"Even on the college entrance exam, I didn't fare this badly. At least on the college entrance exam, I could get the basic points. This exam paper is full of tricky multiple-choice questions!"

The cries of agony rose and fell, never ceasing.

Away from the noise, Su Zhou had already come out of the library with his schoolbag on his back, picked up his younger sister, and gone home.

When I got home, my parents had prepared dinner for us in advance because they were working overtime.

Su Zhou heated up Su Xing's food first, watched her finish eating, and then urged her to do her homework before opening the door to his own room.

His gaze immediately fell on the computer in the corner.

The indicator lights on the computer case were still flashing, and the neural network training program he had installed before leaving was still running.

Su Zhou sat down in front of the chair, moved the mouse, and a black terminal window popped up.

He stared at the numbers on the screen, somewhat surprised.

Training progress: 47%.

He originally estimated that it would take more than two months for this fully connected neural network to complete the entire training cycle on a CPU, but now, in terms of speed, the total time has been greatly compressed.

Moreover, the loss rate is decreasing.

The accuracy rate is increasing.

The speed is more than an order of magnitude faster than the previous CPU running a fully connected network!

At this speed, the training volume that previously took more than two months to complete may now be enough in a week. Moreover, the feature extraction capability of convolutional networks far exceeds that of fully connected networks, and the final classification accuracy will likely see a qualitative leap.

He set up a script to automatically shut down the computer if the training runs normally, and to automatically save the model and shut down the computer if an error occurs during the process.

Then I turned off the lights and lay down on the bed.

Before closing his eyes, he suddenly thought of something.

The computer mall owner mentioned the PicNet Large Scale Visual Recognition Challenge.

PicNet is currently the world's largest image classification dataset and competition platform, jointly initiated by several top universities and held annually.

Participating teams need to train their models on the massive image dataset provided by PicNet, and ultimately compete on the accuracy of image classification.

If his convolutional neural network can perform well on its own small dataset, then theoretically it should also perform well on PicNet's large dataset.

Su Zhou rolled over and used his phone to search for PicNet's official website.

The eligibility section clearly states, "Participating teams must register as institutions of higher learning, research institutions, or enterprises; individuals are not eligible to participate independently."

Su Zhou stared at the line of text for three seconds, then put his phone down.

Universities, research institutions, and enterprises.

He's a high school freshman, so he doesn't fit into any of those categories.

I can't go to the competition.

But Su Zhou wasn't too disappointed. The competition was just one way to verify the results, not the only way.

There is another way.

Write a paper!

If his convolutional neural network does indeed achieve groundbreaking results in image classification tasks, then publishing the method and experimental results as an academic paper would have an impact no less than winning a game.

It could even be bigger!