How to use GPT to publish papers! !

It is said that having top conference papers seems to have its own "traffic".

The school recruitment of many big factories has clearly stated that students who have the top will wait for the buff bonus will be given priority, and they can even have a direct interview without a written test!

fce8d511e5bfaf9468693f166077dce0.png

Of course , it’s not just that high-level thesis is required as an endorsement after graduation to enter a large factory . , When going abroad, it is far more hard-core than other information ...

For the novice scientific researchers who have not yet published their first paper and cannot prove their talents through other aspects , the importance of learning how to write papers and publish top conferences is self-evident .

Is it difficult to get a haircut ? In recent years, the number of papers accepted by the major conferences has been increasing year by year , and friends and classmates around me often hear about the success of the conference . It always makes people feel that it is not that difficult to submit a conference !

But it was not so simple until the actual operation stage ~

It may not be possible to write a good paper or even write a paper if you continue to follow your own ideas . To master the method , it is very important to have someone pointing and guiding !

Recently, the popularity of ChatGPT and large-scale models has been high, and major manufacturers are laying out related research one after another. Baidu's Wenxinyiyan was launched soon after, and AIGC's long-term value establishment was established. At this cusp, how do we respond to the future direction of scientific research, and how do we publish papers in this crack?

On April 11, we invited Mr. Kimi, a Ph.D. in computer science from Harbin Institute of Technology, to share with you: How do CSers publish papers in the cracks when GPT-4 and Wenxin Yiyan are released one after another? From the training algorithm of ChatGPT, to how to come up with new ideas, learn the top essay writing routines!

9b54c517da48906206308a53b0fb1661.jpeg

Scan the QR code to book a live broadcast

9f12edd641783aa2fc87832310071fb7.png

( Accompanied with a paper writing package and tools for the top conference )

As a popular traffic superstar, both the industry and academia are actually taking advantage of its popularity. How about the relevant research? At this time, the papers related to this will not only get good attention in the publication, but also in the follow-up job hunting. stepping stone.

For the students who want to post the top conference , the biggest difference between the top conference / high area papers and the third and fourth district papers is the innovation point. When an article is in the hands of the reviewers, usually the first thing to see is the innovation point of the article. , and then the rest of the paper . It is also often said that it is more popular to use novelty + experimental results to distinguish the quality of papers in the review of top conference papers.

In fact, there are also "routines" to follow in the innovation points of the paper, and the advice from others may play a role of four or two thousand dollars.

The CSer who is still bald for innovation , and the scientific research party who is still worrying about how to write a good paper , must need the experience and guidance from the authors and reviewers of top conferences .

It is very likely that you have been stuck at a certain point for a long time , and you can easily solve it after chatting with the academic seniors .

Scan the QR code to get free pointers on the innovative points of the top conference papers

Limited to the first 100 fans

4ee37498c1c79e84583f162758c41015.png

( Accompanied with a paper writing package and tools for the top conference )

Your possible questions: 1. You have selected topics and haven’t started writing yet

2. There is an idea that cannot be verified/the verification process frequently makes mistakes

3. There is no way to start literature research

4. Unable to determine the research direction and come up with ideas

Guidance you can get:

1. Customized planning guidance for the whole process of scientific research

2. Scientific research experience guidance and pit avoidance guidance

3. According to your foundation and needs, tailor-made scientific research papers

4. Full guidance from topic selection to idea, experiment, data set, writing, submission and revision

How to get an idea that meets the top level?

The suggestion of the big guys is to pay attention to the major conferences in a timely manner , look at their accepted list regularly , and roughly read the work that is closely related to your own direction ( including cross - direction ); then focus on marking the ones you are interested in, preferably Opened up work on code and datasets to reproduce .

Through such steps, determine the subdivision field of the idea you want to do, and then pick out the most relevant work. At this time, only four or five articles are usually screened out , and then intensively read and think about these questions :

What's wrong with this direction ?

Do you have a better way to improve and improve?

In particular , it should be noted that the ideas of ideas and innovations cannot be limited , and papers in other directions that have been read in the past can be used as a reference at this time to improve novelty. After all, under the current review system , most of the people who can review their own papers are their peers. If they only focus on their own field, it is difficult for reviewers to like this job.

Scan the QR code to get free research dry goods from the boss

Limited to the first 100 fans

13746439d02b96c49e0ac9a97befa3f0.png

(Accompanied with a paper writing package and tools for the top conference)

Of course, in addition to innovations and ideas , there are many aspects that need to be paid attention to in the process of writing a thesis . Many beginners do not really know how to write a thesis .

How to write each part of the top conference paper ? Listen to the suggestions of reviewers who have many top-level reviews——

17faa5cf54b542eff1a320ce95b87178.png

1. Abstract

This part is a summary of your entire work. It needs to reflect the essence of the paper without being too lengthy. You can refine and summarize each paragraph of the Introduction part into one sentence, and then connect them together to form an Abstract.

2. Introduction

The important thing in this part is to tell a "good story" so that reviewers can follow your logic and ideas and agree with your work. It is best to include a graphical example that can illustrate the research problem or content to help reviewers people understand.

It can be written in the structure of "five paragraphs"——

Background introduction + problems existing in existing methods + own solutions + ️experiments and results + work contribution statement (the work is to improve a certain scheme of an existing research idea or branch), or background introduction + other research branches Summary + ️ own method + ️ experiments and results + work contribution statement (the work proposes a new framework or structure)

三、RelatedWork/Background

You need to write 2-3 branches most related to your own research content, and each branch must have a subsection, pay attention to the summary, and explain the difference between your method and the previous method.

4. Method

Method should correspond to the contribution statement in Introduction, introduce your method in detail, and highlight the core parts and modules of your work. In addition, a "good-looking" main image is very important, pay attention to color matching, emoticons, legends, etc. Basically, you can roughly understand the process or framework by looking at the picture.

Reasonable use of formulas should be used when writing a Method. Do not pile up too much, which will be counterproductive. Every formula must have value and explanation.

5. Experiments

Contains three parts: ExperimentSettings, including: datasets, data statistics; implementation details of models, frameworks, and methods; introduction of comparison methods and baselines; evaluation process and indicators.

ExperimentResults, that is, the main experimental results reflected by indicators, 1-2 tables and analysis.

FurtherAnalysis, which can include: ablation experiments, case studies, visualizations, A/B tests, etc. This section can also be used as an independent section.

六、Conclusion and Future Work

Summarize the methods and experimental conclusions, and if possible, propose 1-2 directions for future improvement.

19ddd92ae632b5e9b399f37ca4b37031.png

Scan the QR code to get top conference paper writing dry goods for free

Limited to the first 100 fans

180716a2e4d9054674c35b1224d4835f.png

(Accompanied with a paper writing package and tools for the top conference)

What kind of mentor do you want?

1. Rich experience in scientific research, able to answer questions in a timely manner

2. Rich experience in thesis writing, giving guidance on thesis writing

3. Have a shining and strong endorsement

Our tutors:

1. Hold multiple high-level papers and be the first author of the top papers

2. Top reviewers

3. Doctors, post-doctors and other scientific research practitioners in QS Top universities

Welfare at the end of the article


As a researcher who is busy with papers every day, the editor knows that everyone must need some information. Therefore, the editor has carefully compiled a gift package of over 1T AI conference papers! Contains the latest top conference papers, books and other materials, as well as English thesis writing guidance nanny-level materials, from literature reading to thesis writing , all organized for you ~

ab7cc3d61b10a906b495c52fd609184c.png

a55f68ffe5e2fb388039f2a4d01f3221.png

cbf25feb153920ec062052d5a3f9fc29.png

32276ba080df4edba848e5ed03e0f8d8.png

Scan the code to get course materials for free↑

-END-

Guess you like

Origin blog.csdn.net/cainiao_python/article/details/130002462