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

Competition results can only prove that the model scores high, while a high-quality paper can fully explain the methodology, experimental design and theoretical analysis, allowing researchers around the world to reproduce and improve your work.

Only in this way can the group task be completed.

Su Zhou lay in bed, her mind racing.

Where should I submit my paper?

I did some research, and there are currently three top academic conferences in the field of computer vision: CVPR, ICCV, and ECCV.

CVPR is widely recognized as the premier conference in the field of computer vision. It is held every June, with the deadline for submissions usually around November of the previous year.

It's early October now, and there's about a month left until the submission deadline.

Time is tight, but it's not impossible.

But the prerequisite is that he has to write the paper first.

While there was still time, Su Zhou searched for CVPR's submission guidelines.

Paper format requirements: IEEE two-column layout, main text no more than eight pages, unlimited references, abstract, introduction, review of relevant work, method description, etc.

As Su Zhou read through each point, he realized that writing a thesis was much more difficult than he had imagined.

He did all the research and experimentation himself, so writing about it should be easy for him. But the difficulty lies in the norms and skills of academic writing itself.

How to write a high-quality abstract? Which references should be cited in a review of related work? How to design a comparative experiment in the experimental section to make it convincing? What are the considerations for the layout of figures and tables?

It's a huge problem; he's never formally studied any of these things.

He opened his browser and searched for "how difficult is it to write a thesis", and the results almost made him burst out laughing.

On various forums, the cries of despair from graduate students rose and fell, making it a veritable large-scale display of their mental state.

"My supervisor asked me to decide on a research topic within a month. I spent three weeks just reading literature, and I still haven't figured out a research topic. But I've already lost my hair."

"When I got to the Related Work section, I realized that what I thought was an innovative point had already been done by someone else three years ago. My blood pressure spiked on the spot."

"I revised the abstract twenty times. My supervisor said it read like a product manual. I asked him what it should read like, and he said it should read like a research paper. I'm really grateful to him; he was very helpful."

"My advisor's comment on my thesis was that it was written like 'Dream of the Red Chamber,' full of absurd nonsense!"

"Don't ask me about the progress of my thesis. If you ask, I'll say I'm writing it. If you ask again, I'll say I've been writing it the whole time, but actually I haven't written a single word."

"My roommate received the reviewer's comments. Reviewer 2 suggested that he change his direction and redo the experiment. That night, he treated us to a farewell dinner and said he was preparing to extend his graduation."

Su Zhou spent half an hour browsing through the posts, and the more he read, the more disheartened he became.

These are people who have been seriously studying for several years as graduate students, and even with their supervisors providing hands-on guidance, writing a thesis is still so painful.

He was a high school student with no tutor, no research group, and he only learned about the format of academic writing today.

Writing a thesis is indeed a completely different matter from writing code.

Code is all about running, while academic papers are all about impressing reviewers.

Su Zhou rolled over in bed and stared blankly at the ceiling.

Suddenly, he thought of a place.

Group 42.

Almost everyone in that group has published world-changing papers; when it comes to academic writing, they are absolutely pioneers.

Su Zhou picked up her phone and opened group 42.

The group chat is buzzing.

Newton and Leibniz are arguing again.

The incident began when Leibniz presented a differential equation, saying, "This problem is very elegant when expressed using my notation system," to which Newton immediately replied, "The fluxion number method can also solve this problem, and it is more physically intuitive."

The two exchanged barbs, creating a tense atmosphere.

Su Zhou organized his thoughts.

He couldn't just say "I can't write papers," that would be too beneath him.

Su Zhou hesitated for a moment, then typed:

[Eagle Warrior: Everyone, I have a student who is currently trying to write his first academic paper. His experimental data is very good, but he has absolutely no experience in academic writing. I've posted the submission guidelines here; could you please give him some pointers so I can pass them on to him?]

[The Warrior Who Endures the Fly: [Image]]

The image is a screenshot of the CVPR submission guidelines, which includes requirements for sections such as IEEE two-column layout, eight-page main text limit, abstract, introduction, related work, methodological description, and experimental analysis.

After the message was sent, Newton and Leibniz simultaneously stopped arguing.

[Newton: Your student? Interesting. When I was writing *Mathematical Principles of Natural Philosophy*, I spent two years revising it repeatedly, just to ensure that the derivation of every proposition was flawless. I wanted to tell your students that the core of a paper lies not in the elegance of its language, but in the completeness of its logical chain.]

Leibniz: It's rare for me to agree with Isaac, but I'd like to add that notation and notation are very important. A good notation system allows readers to understand what you're doing at a glance. Variable names should be consistent; don't use 'x' one minute and 'X' the next—that's torture for the reader.

[Newton: I've already done what you're saying.]

[Leibniz: What has your fluxion number notation accomplished? Readers need to learn your proprietary notation before they can understand your papers. Is that what you call achieving something?]

Just as the two were about to start arguing again, Turing stepped in.

[Turing: @AoYingWarrior, I'm quite interested in the submission guidelines you outlined. The eight-page limit is actually a good constraint; it allows authors to focus on their core contributions and remove all unnecessary information.]

[Turing suggested that your students first answer three questions: What problem does this paper aim to solve? Why is this problem important? What is the essential difference between your method and existing methods? If they can answer these three questions clearly in three sentences, the paper is already halfway to success.]

[Shannon: Let me talk about the techniques of writing itself. The biggest difference between academic writing and literary writing is that the goal of academic writing is to maximize the efficiency of information transmission, rather than to create aesthetic appeal.]

[Shannon: The abstract is a compressed code of the entire paper. A good abstract should accomplish four things within 150 to 250 words: explain the problem, explain the methods, explain the results, and explain the significance. After reading the abstract, the reader should know whether the paper is worth reading further.]

[Shannon: Also, the introduction should be like a funnel, gradually focusing from the broad research background to the specific problem you want to solve. Don't jump right in with technical details, as the reviewers may not be experts in your specific subfield.]

Shannon posted five messages in one go, each hitting the nail on the head.

Su Zhou quickly tapped the screen to take a screenshot, afraid of missing any detail.

Shannon truly deserves his title as the father of information theory; even when teaching people how to write papers, he considered it from the perspective of information transmission efficiency.

[Turing: Agreed. Also, your students need to be prepared to face questions from reviewers. Reviewers are usually very critical; they will look for flaws from every angle. Therefore, when writing the paper, you should anticipate the reviewers' possible questions and provide answers in advance in the paper.]

[Shannon: Tell your students to remember this: Good papers aren't written, they're revised. The first draft doesn't need to be perfect; just fill in the content first, then polish it repeatedly.]

[Newton: Indeed, my "Principia" has been revised countless times. Of course, if it weren't for some people plagiarizing my work, I could have revised it more easily.]

Leibniz: ???

[Leibniz: Isaac, do you have to find a chance to make sarcastic remarks about me every day? Aren't you tired of it?]

Su Zhou smiled, knowing he couldn't persuade the two of them, so he left the group chat, leaving them to continue arguing.

He put away his phone and sat down at his desk with a serious expression.

Shannon's framework was so clear; he practically laid out the skeleton of the paper step by step.

What he needs to do now is stuff meat into the skeleton!