adkid-zephyr/oral-paper-skill

** I’ve distilled 883 final oral papers, distilling the commonalities and learning-worthy practices that help you apply to your own ideas and papers. **

Oral paper, which refers to papers selected by academic conferences for oral reporting, is usually given attention for research contributions, innovativeness or influence, and is an example of excellent papers worth learning.

English README . Complete Skill . Chinese Phrasing

Why do this, Skill

I think there’s a reason that oral paper can be oral paper, and there’s a reason to study and learn. If you want to understand more quickly how to make and write a good paper, it is a very direct learning material.

The first step in receiving systematic scientific training is often the reading of classic articles. The previous sessions of the Conference, oral, and other papers of high quality and gold, are also common examples.

ICLR, ICLR, ICLR, ICMR, ICML, NeuroPS, do the same. It’s an inspiration, it’s a way to get an A into a B scene, or it’s a local optimization, and it’s a new paper.

A lot of people published papers through this road. But what exactly is left of this kind of learning, if only to learn at the end A+B and a little bit more? Portfolio and improvement can certainly be of value, and the key is whether we understand why the issue is important, why the innovation is established, and how the evidence supports the conclusion.

With AI, I think the process can be more systematic: quickly collating and better dissertation, distilling useful commonalities, then putting these practices in their own ideaa, helping to learn scientific research and writing.

I’ve been thinking about what to share since the last video I shared [T0] Skill raised a few hundred. I’ve built up and hid a lot of good Skill, and this time I’ve made Oral PaperSmart a second project in this series.

** My goal is to use the official Oral entries of the ICLR, ICML, NeuurIPS and the related excellent papers of the nearly two completed cycles as learning materials, extracted, compared and integrated, to form a reusable methodology. **

The term “distillation” here refers to the extraction of knowledge: the compilation of specific practices, conditions of application and examples from the paper, in the form of Skill. The round is extracted from the summaries, which are then compared, verified and synthesized; recommendations that involve the body or chart require additional reading of the corresponding material.

It does two things for you

#1. Find something to improve against good papers

Select the appropriate example, depending on your problems and the type of paper, to explain how it expresses its contribution, organizational evidence and then indicate how your contribution can be modified.

Each of the most important recommendations contains as much as possible:** the original thesis approach and the source why it is relevant to you to review the status of your writing and to make specific modifications. **

For example, in the face of a paper advocating lower costing, it is possible to check whether it gives end-to-end cost comparisons of the same quality. This is an indication of the form of the recommendation; when the actual attribution is given to a paper, the corresponding text needs to be read.

#2. Learn good practices and guide your reset

Clarify the approach that is worth learning: how the introduction goes forward, how the innovation points are distinguished from the previous, how the map is presented, and how key experiments exclude other interpretations。