🥇 copyright: this article by [Moscience AI] original, first, you big guy, please refer to 🎉 statement: as one of the bloggers in AI field, ❤️ lives up to your time and your ❤️ ❤️ if the article is helpful to you, welcome to three links
In daily life, I helped my boss review a lot of papers and summed up some experiences to share with you who are on the way
Computer vision, superfraction reconstruction, image repair, target detection, model deployment, etc
📕 Eight tips for essay writing
The purpose of writing a paper is to be published. When publishing a paper, you need to write it according to the routine of the paper, otherwise the reviewer will reject the paper as “unprofessional”. It is perhaps the most painful thing to be rejected for not following the routine when you have worked so hard for your scientific research achievements.
1, all things difficult Title – topic selection
Title and abstract are the contents that can be seen in the retrieval of the paper, which directly determine the reader’s desire to read on, so it is particularly important. The principle of topic selection: the direction of topic selection is as small as possible, the least vocabulary is the most comprehensive expression of the topic of the paper, and strive to be accurate, concise and clear, to avoid a look is generalized.
One way to speculate: Keyword search according to the paper published articles in recent years, be modified in the light of their own use, you can also download the corresponding articles, as their writing template (if they are not thrown a journal before, it is very important to choose writing template, on the one hand, allows you to avoid stepping on many unskilled pit, on the other hand can provide guidance for the layout of the thesis, Including some grammar and expression can also be more idiomatic).
2. Write the main article
After choosing the topic, first of all, download the paper from which the topic comes and use it as a writing template. Use the outline of this paper as the skeleton of your paper, but the flesh and blood needs to come from other places. If it is a related working chapter, find the content of relevant literature to fill; If it is the method section, refer to the template article according to the actual situation, modify and use it.
3, in this paper,
The purpose of the abstract is to let readers know what the paper is about after reading it, mainly including research significance, methods and results. In popular terms: why do it, how do it, what are the results.
It is usually written in the third person. First, one or two sentences summarize the meaning of the topic and introduce the topic, that is, the problem to be solved. Then, in order to introduce the research methods, a general description of each step; Finally, the experimental results and significance are given. The experimental results should be rigorous, such as correlation of 86.9%, accuracy of 90%, etc.
Example: In order to solve the problems of long design cycle and time-consuming and laborious sample making of colored fabric, a simulation method of realistic colored fabric with color spinning image was proposed. Firstly, the color yarn image was collected, and the yarn body was obtained by threshold segmentation and morphological processing. The upper and lower boundaries and center lines of the yarn body were obtained, and then the main part of the original yarn image was obtained. Then, according to elliptic model and sinusoidal model, the yarn body image is processed to get the shape of yarn in the two-dimensional fabric surface. Finally, the covering relationship of warp and weft yarns was changed according to the color yarn cycle and fabric structure transformation model, and the realistic striped and lattice colored fabrics were simulated. The simulation results show that the proposed algorithm can simulate the weaving process of different kinds of colored fabrics, reflect the appearance effect of the fabric truly, and can adjust the fabric structure and color yarn cycle parameters, which improves the authenticity and adaptability of the existing simulation algorithm.
Note:
(1) the research significance, the need to solve the problem not only solve the problems in the research of predecessors, such as fabric simulation is in order to solve the problem of proofing time-consuming, then other fabric simulation method can solve, to be on the faults of others to build their own innovations, here you should write “color fabric simulation effect is dull, lack of sense of reality, Problem of high model complexity “;
(2) in the experimental method, only need to explain what is used, what is the purpose of good, do not need to give specific parameters, charts and references, etc., generally there are word requirements;
(3) In the experimental results, do not use “higher”, “better” and “about” and other function words, to have specific experimental results. Do not appear self-evaluation words, such as “first proposed”, “of great significance or practical value” and so on.
(4) the abbreviation is stated when it appears for the first time. But when it first appears in the text, it should also be stated.
4, introduce
The purpose is to explain the relevant research background and significance and the organizational structure of this paper. The form can be: first throw out a question, this problem is very important, the summary in order to solve this problem put forward one, two, three methods, but one, two, three methods and what are the shortcomings, the middle can be in a progressive relationship to typesetting. Then I will introduce my own method, the characteristics and advantages of our method, and the final result summary, and then I can introduce the layout of the following article.
5. Related work
That is to study the current situation, what those people are doing, what their value is, what their flaws are. Different from the introduction section, this section is the specific work of a specific person, and the introduction section is more like a summary statement after classification of methods.
Here, you need to read and summarize a large number of literature, and explain the shortcomings of previous studies or areas needing improvement in appropriate parts, and present the innovation of this paper in comparison, that is, explain why you want to do this work.
First of all, you need to introduce your work below. According to the shortcomings mentioned above, you need to explain the motivation for this work, what research you did in this paper, and how you did each step. It is corresponding to the steps in the abstract, and there is no need to write a conclusion. Finally, list your innovations in this article in order.
Note that:
1. This part needs to classify the work of predecessors (there are many classification standards, such as classification by method, classification by time, etc.)
2, each job do not introduce too detailed, one or two sentences on the line, as far as possible to the original expression. Explain the advantages and disadvantages of each previous work.
Pay attention to the point
(1) Journal literature with high impact factors should be used in the current research situation, and of course, the literature of the target journal should be used less or no graduation thesis, conference papers and publications;
(2) The research focus should be on the literature in recent years, and the old literature should not be used, unless it is at the level of ancestors and masters, which is very important for the article;
(3) The description of the research steps is an overview of the material and method steps, which can be similar to the abstract, but should not be repeated too much.
6. The proposed method
It is mainly to introduce your work. First of all, it needs to be clear that the summary of your work has more or less appeared in the abstract, introduction and related work. The steps of each part should be consistent to ensure that the main line is clear and the logic is clear.
(1) Firstly, the method or model used should be simply introduced, as long as the content can explain the problem. There is no need to repeat the content in previous literature, such as principles, definitions or formulas.
(2) Each paragraph of the experimental steps should have a topic, explain what was done in this step, what was the purpose and what results were obtained, and make it clear.
(3) Use a simple short game on the language, try to avoid the use of long sentences, easy to produce ambiguity. But the terminology must be standard.
7. Experiment and result analysis
This part is mainly about the experimental support of your method, your innovation points, performance gains need to be verified in this part of the experiment. So the most important thing is experimental design:
(1) Correlation of experiments. To design experiments around the topic of your own paper, you can refer to the experimental designs of other researchers.
(2) There is a general principle in this part that the experimental results that are recognized must be repeatable, that is, other researchers can reproduce the results in the experiment according to your description. [Big guy, top meeting necessary]
(3) Comparative experiments are indispensable. Performance gains can’t be a one-size-fits-all story. You have to compare it to other people’s methods and let the numbers speak for themselves. And finally, and most importantly, what are you trying to prove with this experiment, what conclusions do you draw, and do those conclusions support your argument?
At the beginning of this chapter, experimental data sets will be introduced first, and public data sets will be selected as far as possible. Secondly, the experimental steps and conclusions are described.
Note:
(1) Well-grounded: The setting of each parameter should be based on, such as the threshold value, etc., even if you see it from others’ articles, you should also mark references, but it is recommended to give the best parameters through experiments, others’ parameters may not be applicable to your experiment;
(2) Formula editing: It is suggested to write in MathType, and the insertion formula in Word is not standard, let alone screenshots from others’ papers;
(3) Chart drawing: General reviewers like to browse the chart quality first to judge the quality of the paper. Photos of experimental equipment or experimental samples can be taken by themselves and are more favored by reviewers. The graph must be an editable vector graph. It is recommended to use PPT or other software to draw the graph illustrating the experimental method, and Origin or Excel can be used to draw the graph of data analysis. For tables, just insert table drawing. A chart is a visual description. The chart appears after the text. Remember that the content of the chart cannot be repeated in the form of words, just explain what is reflected.
(4) Symbols: For variables, use italics. Variables can be no more than two letters and can use upper and lower indices. Variable symbols need to be explained when they first appear. Other specific requirements depend on different periods.
8, conclusion
The conclusion is a summary of the paper, which is mainly a brief summary of the experimental scheme, what data and phenomena are obtained or observed from the experiment in the text, which problems are solved in turn, and the similarities and differences with previous studies. It should be noted that the experimental scheme here is not as detailed as in the abstract and preface. What is important is the result and significance. Finally, the prospect is given, that is, suggestions for further study of this subject.
📗 Eight tips on paper writing
Number 8, too true
1. Writing a paper is a system engineering
It is different from article writing, so from the first day we are ready to start writing, we need to have an overall thinking and planning, such as document management, document indexing, data management, tables and images management, etc. It is recommended to start with a folder that includes subfolders for each aspect, categories, system administration, and multitasking.
Writing is not accomplished overnight
Many people are miserable when they write a paper and write half a page one day and hundreds of words the next. Before you write, think about it, clear your mind, and start writing. Sometimes it is not unusual to finish a paper in a day or two.
Don’t expect to write everything in one draft
The most important part of writing a paper is the first draft, the hardest part is starting from scratch. It is not difficult to revise the first draft after it comes out, no matter how sharp and tedious the revision opinions are. As the saying goes, the hard work makes the steel. The same is true when writing a paper. The final draft is completely different from the first draft. Therefore, in the first draft, we need to be fast, not fine.
4. Don’t rush to generate a list of documents for a directory
Because once the document list is inserted, it will always be updated again and again as it changes, sometimes causing Word to crash, resulting in unsaved or missing files.
5. There is a difference between writing a paper and writing an article
For a paper, refinement is not the most important requirement, details are the most important. For articles, brevity is the most important, because magazines usually have a word limit, and concise words are more favored by magazines. So when we write papers, we have to try to be very detailed in our first draft and then revise it a little bit.
6, In the process of writing a paper, there are often some small problems
For example, why the format of the table of contents is always wrong, why the “update page only” option of the table of contents is not available, how to insert the page number on the left side of the horizontal page, how to customize the page number of the footer, and how to distinguish between centerfold interrupts and fragment interrupts. . Remember, if you encounter a similar problem, identify it and resolve it as soon as possible. It’s not the design of Word that’s wrong. Then you’re not good with him. Most of the time, the problem is solved through an Internet search. Therefore, these questions will not bother you later.
7. Timely communicate with the supervisor about the writing progress and requirements of the thesis
Every tutor has different habits. Final quality control rests with the mentor, so communication is of Paramount importance.
Take each paper as a starting point for your rigorous scientific research
Everyone knows that a published paper is rarely read, and even if it ends up in the library, it may never be read. However, the harvest in the process of writing a paper is to write an article can not be compared, the opportunity to write a paper is less, that is to say, the system to write a paper is less opportunity. When you have more opportunities to write articles, you will find that articles are actually easy to write and follow the same pattern. Therefore, if you really want to become a serious scientist in the future, please write your papers carefully, even if no one will read them.
The original address is as follows
https://positive.blog.csdn.net/article/details/121719598
Copy the code📘 more effective tutorials
The content of this blog is difficult to generalize, if there are inappropriate places, welcome to the comment area criticism and correction
For those who are about to enter the line of computer vision, Murray here recommend the collection of dry goods blog is as follows
-
🎉 deep learning model training reasoning — basic environment building recommended blog review order — [1024 special issue]
-
Juejin. Cn/post / 703181…
-
Thank you for your thumbs up and comments to encourage bloggers to create every minute seriously
📙 Friendly reminder
- We’re 0-6 months away from the essay contest
- Every year from March to June, check the price – index level mutation
- Surplus food is limited, followers of bloggers should know how to stock up in advance wow
- As long as within the validity period, it is recommended that each young partner who is about to graduate can stock two or three copies
- There is an object tun double, a good man is you
- No new(), no new(), then you can try
buy one lipstick or basketball - It’s a lot of money. Murray’s not responsible here
❤️ Life is too short, welcome to learn AI with Murray
🎉 as one of the bloggers with the most dry goods in the field of AI, ❤️ lives up to his time and qing ❤️ ❤️ If the article was helpful to you, don’t forget to comment encouragingly
