Now that we have completed working on our project. We can now extend to write a paper if the project involves proper techniques and results. At this point, the mentor helps in getting started with writing the paper.
3. Related Works.
4. System Architecture.
6. Results and analysis.
There are several stages in writing a paper. The first part of a paper is the abstract. The abstract contains a brief idea of the project. It gives an idea of various methods, software used, and various types of results taken while developing the project. This part gives an idea of what the paper contains.
With the help of this many readers can pick up the proper papers useful to them. There is a small section below the abstract for specifying keywords in the project which is about 2 lines. The abstract is written in bold letters.
This section introduces the readers to the project. It discusses the problem statement, its importance, its impact on society. On further reading, it will introduce the readers to various techniques and implementations involved in their work.
During the idea development phase whatever papers we have referred to would be listed out in this section. The author should give a small summary of what the referred paper conveys. By this we are able to cite various papers, this in fact helps a lot.
It helps the actual authors whose work has been referred to. The more the number of citations the more would be the value to the paper. Through this section, the authors of the paper can recommend their readers about what all the fascinating work is happening in the world. In the future, someone may cite the present paper through which the present work would get value.
This section gives the top-level view of the entire workflow of the work. This section contains the block diagram of the project. The complete pipeline will be discussed in this part. If the project involves any mathematical equations then they are presented here. For example in a machine learning project, The details such as data filtering, data augmentation techniques, type of ml techniques used will be discussed.
In this part, since we discuss the overall procedure, it helps the readers to relate our work with their work and let them empower their idea.
In one of my previous projects in deep learning, this part helped me in navigating to various open-source datasets. From this section, I have gained a lot of understanding of the higher-level view of the deep learning pipeline for large scale projects such as OCR, image captioning, etc.
In the previous section, we will give a top-level view of our project. In this section, we would go further in-depth about the main part of the project work. Many great innovations described the theoretical working of their works in this section. In machine learning, various theoretical concepts are first described here and later were proved in the results section.
For any hardware project, we would discuss the main hardware architecture, for machine learning projects we would discuss the algorithms used and how they contribute to achieving greater results.
This section lets the reader go a step further in understanding why the proposed architecture is best suitable for getting greater results. For example, in image classification, this section helps a lot in choosing the best architecture for the task.
Results and analysis
This section contains the results and analysis of the project work. It is always good to have various parameters for obtaining the results of the project. If we have more parameters to judge our prototype, then we would be able to see our prototype from various angles. This helps us to check the performance of our prototype. Each parameter would describe the prototype in various ways. If there is any small error in modeling then that may be pointed out here. This helps in improving the prototype further.
In the paper, this section holds a greater weightage. We can convey our results in the form of tables, graphs, images. It would be good if we can provide as many details as possible regarding our work. This helps the reader to relate methods and their outcomes. If the results obtained are best, then the reader would try to replicate the same. This section is proof of our description in the previous sections.
In the past in the field of deep learning various models such as inception, resnets, and other great models became so popular based on this section. This section proved many inventions to be the best.
This section contains the last words regarding the author’s work. If the author has compared several methods then he can give a conclusion of the best performing method. The author can describe the future scope of their work. This section contains the entire story of the project in small words and will end in a small paragraph.
This is the last section of the entire paper. The author needs to mention all these references in this section. In the related works section, the author will mention the work done by various pioneers in small sentences. In this section, he needs to cite their paper. Through this, the reader can able to approach their papers. In general, there should be 15 references.
This brings to the end of the discussion.
Thank you readers,
I hope that this article will help you in building your own projects and papers.
I am an undergraduate student. So whatever I have mentioned in this blog is what I have observed from my little experiences. If there are any mistakes, then please mention them in the comments.
Vedha Krishna Yarasuri