AI Paper Report
[Thanks to Yeganeh Jalalipour for authoring much of this.]
Choose a paper of interest in AI and talk about it.
What To Do
- Summarize the paper.
- You should summarize the paper in your own words.
- You can use this to help yourself to add more
information to the summary:
- Is the title well-suited to convey the research work?
- Is the abstract and conclusion are suited for the body of the paper?
- Does the abstract give you enough information to realize if it was a paper that you find interesting? Do the abstract and introduction give enough information?
- What were the assumptions authors used in their work?
- What dataset did the authors use? Does this dataset work to support their conclusions? Is it enough data and of good quality?
- Explain the problem that the authors were trying to solve.
- Extract the authors' key points and contributions to the AI community. Look at the impact of the paper on AI research and in the real world.
- What software did the authors implement? What would it take to replicate this work?
- What things are missing from this paper?
- [Optional] Implement or replicate work from the paper.
Overall Paper Organization
Your report should consist of the following elements, in order.
"A report on paper title by paper authors"
"Abstract": a short paragraph summarizing your report
"Paper Summary" section
"Research Analysis" section
"Reflections" section, describing what you got out of the paper
"Bibliography" — must at least include a full citation of the paper you are reviewing
For each text section, write a few paragraphs. Maximum length of this report is five pages typeset reasonably — three pages would be a better target. There is no minimum length, but you must adequately discuss the topics.
How To Submit
Your should submit a ZIP archive containing two files:
A PDF version of your report — only PDF will be accepted, no Word documents, Google docs links, etc — with the filename
If possible and legal, also submit the PDF of the article you are reporting on. Name it
Be sure to include full citation information in your report, including a URL where possible, so that we can find the paper you are reporting on and understand where it came from.
Where To Find Papers
You can search for topic keywords, author names, etc. (You can access more research papers if you are connected to PSU's VPN.)
You can also log into the PSU library and look for the name of the paper you have in order to download the free version of the paper if it is available through PSU or affiliated libraries.
This is an open-access archive of research papers. Some researchers publish their research here before or after publishing them elsewhere.
The Computer Vision Foundation has mostly computer-vision-related work, but you can find many papers here. This includes papers that were accepted to top conferences in computer vision and machine learning such as CVPR and ICCV.
Advances in Neural Information Processing Systems.
Being connected to PSU's VPN or wifi, you can access papers in the following journals as well:
You can also just use a search engine — Google or whatever — to look for AI-related journals and papers.